Uploaded by Usha B S

182019007047 v1 220318

advertisement
Generated on 10-06-2019 12:06:38 PM
Advanced Machine Learning Algorithms for
DIGITAL SOIL MAPPING AND SOIL FERTILITY DEGRADATION
ASSESSMENT USING HYPER SPECTRAL, SATELLITE IMAGES AND
FIELD SPECTROMETER
.
Reference No. : 182019007047
Saved By : Dr. SINGH VIPULA
Saved Date : 08-Jun-2019
Ref No. : 182019007047 | Page 1 of 33
PROPOSAL DETAILS
Dr. SINGH VIPULA
vipulasingh@yahoo.com
PROFESSOR AND HOD(ELECTRONICS AND
COMMUNICATION)
RNS Institute of Technology
Rns institute of technology,
dr.vishnuvardhan road, rajarajeshwarinagar, channasandra,
bengaluru, karnataka 560098, Bangalore urban district, Karnataka560098
Technical Details :
Scheme :
Research Area :
Duration :
Date of Birth :
Nationality :
Core Research Grant
Electrical Electronics & Computer Engineering (Engineering Sciences)
36 Months
Contact No :
+918105917500
10-Sep-1971
Total Cost (INR) :
22,31,000
INDIAN
Is PI from National Laboratory/Research Institution ?
No
Project Summary :
B. S. Das et al [11] claim that most of the soil quality studies in India are done using
chemical analysis which is very labor intensive and time consuming. Actual soil samples
have to be physically collected for every 100 meters from the entire area of study. The soil
composition of these samples are done in the laboratory by chemical analysis and
laboratory reflectance spectroscopy. The findings of the chemical analysis are manually
marked on the maps. Recently SUJALA 3 project has been completed using the same
approach by Government of India for Karnataka region with World Bank funding. Digital
soil mapping (DSM) is a cost-effective and time efficient method for generating soil maps.
Hence an automated approach to digitally map soil degradation is the need of the hour.
Hyperspectral and satellite images contain the signature of soil properties which can be
extracted using sophisticated algorithms. In comparison to laboratory reflectance
spectroscopy and chemical analysis of soil, the use of digital images is less common. One
main reason for the limited number of studies with image processing techniques are
restricted data availability, corrections of atmospheric effects, georectification of the data
etc. Hyperspectral and satellite images can be obtained from ISRO/open source for study
purpose. The signal-to-noise ratio of available hyperspectral and satellite image data is
relatively low compared to laboratory data. To extract soil properties from these images to
match the ground truth is a challenge. Literature survey suggests that the above
mentioned approaches are in an infancy state in India. Few studies carried out have
shown promising results which is a motivation for us to take up this research work. Digital
soil mapping and soil fertility degradation assessment studies have to be carried out for
the entire nation. This can lead to automatic suggestions for farmers for growing alternate
crops, using appropriate amount of specific fertilizer for a particular crop based on soil
properties. Hence we are proposing a comprehensive machine learning approach for
digital soil mapping and soil fertility degradation assessment using hyperspectral, satellite
images and field spectrometer.
Objectives :
• Soil degradation, problems and degree of degradation
• Soil Salinity mapping
• Soil classification and digital mapping
Keywords :
Soil salinity analysis, digital soil mapping, hyperspectral images, Machine Learning
Algorithms, satellite images
Expected Output and Outcome of the proposal :
Digital map of soil properties of area considered, soil salinity map, degree of soil
degradation.
SNo. CO-PI Details
USHA BS
1
bsusha@gmail.com
Assistant Professor(Electronics and Communication Engineering)
R N S Institute of Technology
RNSIT, Dr Vishnuvardhan Road, Channasandra Post,Bangalore, KARNATAKA,
BANGALORE
URBAN
D I S T R I C T
Ref No. : 182019007047 | Page 2 of 33
Other Technical Details
1. Origin of the Proposal:
Advanced Machine Learning Algorithms for
DIGITAL SOIL MAPPING AND SOIL FERTILITY DEGRADATION ASSESSMENT
USING HYPERSPECTRAL, SATELLITE IMAGES AND FIELD SPECTROMETER
Environmental quality is essential for agricultural sustainability of a country. Soil is an essential
national asset and has to be protected for terrestrial living organisms to survive. Water and air
quality of an area is dependent on soil quality of that area. Dynamic properties of the soil changes
due to excessive use of chemical fertilizers. Indiscriminant use of these fertilizers has led to
physical and chemical degradation of soil quality in India and across the world.
Quality of soil can be determined by analyzing the Soil properties. Precise measurement of soil
quality within certain technical and economic constraints is a big challenge. Mineral composition,
Organic matter, soil carbon, moisture, texture has a major influence on soil quality. Organic matter
affects nutrient retention. These indicators must be measured using laboratory analysis to evaluate
the soil quality. Chemical indicators can tell about the soils’ mineral composition, water, nutrients,
clay particles, organic matter as well as levels of soil contaminants. Electrical Conductivity (EC),
Soil Nitrate (NO3-), pH value are some of the prime Indicators that contribute in evaluating soil
quality. Few fertility Indicators are magnesium, phosphorus, calcium, sulfur, nitrogen, potassium,
boron, and zinc and few other minerals. C:N ratio, microbial biomass carbon, soil enzymes, soil
organic matter are Indicators of Organic Matter.
In India, out of 142 million hectare (mh) of irrigated land, about 9 mh which is about 21.4% of
total cultivable land is said to be saline/alkaline. About 80% of the agriculture land is acidic in
nature. Due to continuous usage of land for growing the same crop and excessive irrigation has
led to secondary salinization. Eventually, productivity and area of cultivable land is decreasing
every year. Soil salinity is one of the major causes of soil degradation and is a severe environmental
hazard that has impacted the growth of crops due to excessive use of inorganic fertilizers. Due to
use of excess water/pesticides by the farmers has resulted in storage of excess salt stored in the
soil profile. This salinization problems continue to spread at a fast rate resulting in desertation of
agricultural lands which is a major hazard. There is a direct impact on the agricultural produce of
the state due to increase in soil salinity.
Bellary district in Karnataka is a sitting example of this situation. Water scarcity, land degradation
and poor socio-economic conditions are current problems which have resulted in extreme poverty
and increased farmer suicides, not only in Bellary but across India.
Science-led interventions are the need of the hour for soil and water conservation practices,
productivity enhancement activities, crop diversification for improving socio-economic
conditions. There is also a need to build capacity of the farmers in the region for improving rural
livelihoods through knowledge sharing, alternate occupation and dissemination strategy.
Detection of soil properties in collaboration with Centre for Applied Research on Problematic
Soils (CARPS), Mangalore will help in Soil resources classification and mapping, detection of
Ref No. : 182019007047 | Page 3 of 33
soil acidity and salinity degree. It also enables early detection of the problems of soil and take
necessary step to prevent the spread of soil degradation.
2. Review of status of Research and Development in the subject
2.1International Status:
Zhuo Luoa et al [1] suggested mapping of soil organic material content by using airborne
hyperspectral images. Brightness, coloration, hue, redness and saturation index was obtained using
correlation analysis. Regression model was built using these correlations and the soil organic
material content was mapped. Multivariate regression model with first derivative reflectance of
field spectrums was performed related to soil organic matter. The authors claim that the maps were
76% accurate.
Andreas Kamilaris et al [2] reviewed 40 research papers that used deep learning techniques for
various agricultural and food production challenges. They studied specific models and frameworks
employed, data used. They compared deep learning techniques with existing popular techniques.
The authors concluded that deep learning provides high accuracy, outperforming existing
commonly used image processing techniques.
W. Dean Hively et al [3] collected data from an airborne imaging spectrometer with 400–2450
nm, ∼10nm resolution, 2.5m spatial resolution. They analyzed soil samples for carbon content,
particle size distribution, and 15 ergonomically important elements. The authors used partial least
squares (PLS) reflectance spectra derived from regression model on imagery, to predict analyte
concentrations. Predicted map values were compared with field sample results indicating the
influence of environmental processes on soil properties.
Michael L. Whiting et al [4] developed application to airborne hyperspectral imagery in mapping
crops variation. Water status determination through plant and soil condition is also done. The
authors concluded that precision agriculture is possible through hyperspectral image analysis.
Daniel Vogt et al [5] used soil spectra and chemical laboratory reference measurements to model
carbon, nitrogen, organic matter and carbonate concentrations in soil samples. Classification and
regression tree statistical methods were used to fit to each spectral response.
Eliza E Camaso et al [6] used LIDAR and arial data for spatial analysis and classified N, P, Zn and
S deficiency areas. They also calculated soil salinity with 99% accuracy.
José Padarian et al [7] used convolutional neural network (CNN) model to contextual information
of the landscape and to generate soil maps. CNN models incorporates contextual information
surrounding an observation to significantly improve the prediction accuracy over conventional
NN. They also concluded that CNN model produces less prediction uncertainty and it is able to
predict soil carbon at deeper soil layers more accurately.
Pham Viet Hoa et al [8] used five state-of-the-art machine learning models, Multilayer Perceptron
Neural Networks (MLP-NN), Radial Basis Function Neural Networks (RBF-NN), Gaussian
Processes (GP), support Vector Regression (SVR), and Random Forests (RF) to map soil salinity
Ref No. : 182019007047 | Page 4 of 33
on Synthetic Aperture Radar (SAR) C-band data. They compared the results with soil sample data
collected from field. They concluded that incorporating machine learning methods and Sentinel-1
radar imagery to produce soil map has good accuracy.
Michael Vohland et al [9] compared non-imaging laboratory and airborne imaging spectroscopy
to assess arable soil quality indicators. They studied microbial biomass-C (MBC) and hot waterextractable C (HWEC), organic carbon (OC) and nitrogen (N).
Gholizadeh, A. et al [10] made a study to assess the potential of superspectral Sentinel-2 satellite
for predicting and mapping the soil attributes. The prediction accuracy based on lab spectroscopy,
airborne and spaceborne techniques in the majority of the sites was adequate for SOC and fair for
clay.
In [11] researchers have worked to find a relation between hyperspectral images and ground-based
measured soil features in the affected area. The spectra of soil samples in two affected areas were
compared with hyperspectral images. They have confirmed through their research that
hyperspectral images are useful and an efficient tool for soil contamination mapping.
2.2National Status:
B. S. Das et al [12] studied the preparedness and opportunities for using the Hyperspectral Remote
Sensing approach for soil assessment in India. They studied the availability of data and algorithms
that can be used for the data. Authors also claim that most of the studies in India is done using
chemical analysis which is very labor intensive and time consuming. Hyperspectral image analysis
provides very good alternate to lab test. A joint study undertaken by NBSS&LUP, Nagpur and
SAC, Ahmedabad showed significant negative correlation between soil OC and soil reflectance
data of 139 bands of Hyperion image.
P. Nila Rekha et al [13] conducted hyperspectral analysis to ascertain the salinity status of soil
samples in 2 regions. They used Vertical Electrical Sounding (VES) to test electrical resistivity to
measure soil salinity. They compared the results with hyperspectral image analysis.
K.N. Chaudhari [14] developed a methodology for crop growth monitoring and yield forecasting.
Two different approaches; the forcing approach and the regression approach were used and
compared for spatial wheat yield forecasting. They used dynamic simulation model-WOFOST and
RS derived crop parameters like crop fraction, LAI for crop yield forecasting
The authors in [15] studied land degradation and generated maps for the country and also for
Karnataka.
M. a. E. AbdelRahman et al [16, 17] developed a methodology to classify type, severity and cause
of land degradation for Chamarajnagar dist Karnataka. They have used RS, GIS and GLASOD
approaches and concluded that salinization is occupying 5.19 per cent, Dystrification/(chemical
degradation, leaching) occupying 12.46 per cent, Compaction occupying 14.79 per cent,
Wasteland occupying 4.58 per cent and Water erosion, Loss of topsoil occupying 22.57 per cent,
of the total area.
The authors in [18] studied desertification and land degradation status mapping for India on 1 :
500,000 scale using multi-temporal satellite data. The authors also studied land degradation in
Ref No. : 182019007047 | Page 5 of 33
terms of water erosion, vegetal degradation, wind erosion, salinization/alkalization, water logging
etc. The study reveals that 105.48 mha area of the country is undergoing processes of land
degradation (32.07% of the total geographic area of the country) and area undergoing
desertification is 81.4 mha [18].
2.3 Importance of the proposed project in the context of current status
B. S. Das et al [11] claim that most of the soil quality studies in India are done using chemical
analysis which is very labor intensive and time consuming. Actual soil samples have to be
physically collected for every 100 meters from the entire area of study. The soil composition of
these samples are done in the laboratory by chemical analysis and laboratory reflectance
spectroscopy. The findings of the chemical analysis are manually marked on the maps. Recently
SUJALA 3 project has been completed using the same approach by Government of India for
Karnataka region with World Bank funding.
Digital soil mapping (DSM) is a cost-effective and time efficient method for generating soil maps.
Hence an automated approach to digitally map soil degradation is the need of the hour.
Hyperspectral and satellite images contain the signature of soil properties which can be extracted
using sophisticated algorithms. In comparison to laboratory reflectance spectroscopy and chemical
analysis of soil, the use of digital images is less common. One main reason for the limited number
of studies with image processing techniques are restricted data availability, corrections of
atmospheric effects, georectification of the data etc. Hyperspectral and satellite images can be
obtained from ISRO/open source for study purpose. The signal-to-noise ratio of available
hyperspectral and satellite image data is relatively low compared to laboratory data. To extract soil
properties from these images to match the ground truth is a challenge.
Literature survey suggests that the above mentioned approaches are in an infancy state in India.
Few studies carried out have shown promising results which is a motivation for us to take up this
research work. Digital soil mapping and soil fertility degradation assessment studies have to be
carried out for the entire nation. This can lead to automatic suggestions for farmers for growing
alternate crops, using appropriate amount of specific fertilizer for a particular crop based on soil
properties.
Hence we are proposing a comprehensive machine learning approach for digital soil mapping and
soil fertility degradation assessment using hyperspectral, satellite images and field spectrometer.
The objectives of the proposed research work are
1. Soil degradation, problems and degree of degradation
2. Soil Salinity mapping
3. Soil classification and digital mapping
2.4 If the project is location specific, basis for selection of location be highlighted: NA
3.Work Plan:
3.1Methodology:
We propose to select the most affected districts of Karnataka for digital soil mapping and soil
fertility degradation assessment. Four parallel studies are proposed as shown in fig 1.
Ref No. : 182019007047 | Page 6 of 33
a)
b)
c)
d)
Hyperspectral Image analysis
2D satellite / UAV image analysis
Chemical analysis of soil samples
Spectral analysis of soil samples
STUDY SITE AND DATA COLLECTION
Air dried & grinding
SPECROSCOPIC MEASUREMENTS
Hyper Spectral
Images
Laboratory
Measurement
s
Satellite
Images
MULTIVARIATE CALIBRATION AND CROSS VALIDATION
Spectral
Matching &
Classificatio
n
Machine
Learning
Algorithm &
Classification
Chemical
Analysis
Spectral
Analysis
CORRELATIVE ANALYSIS
Estimation accuracies
achieved on hyper
spectral images, satellite
images & chemical and
spectral analysis
Compare
Clarify
Estimation accuracies
achieved on hyper
spectral images,
satellite images &
chemical and spectral
analysis
Final desertification / Land degradation status map
Fig 1. Work flow of the study
Mineral composition, organic matter, soil moisture and texture are four factors that majorly
influence remote sensing signature of the soil. Soil properties shall be extracted from
Hyperspectral images, 2D images, also from Chemical and spectral analysis. A correlative analysis
will be done for accuracy estimation of parameters extracted from image, chemical and spectral
analysis of the soil samples. Soil degradation assessment report and digital maps of various soil
properties of the study area will be generated after verification with the ground truth.
Ref No. : 182019007047 | Page 7 of 33
Implementation methodology is as follows
I.
Data Collection
II.
Hyperspectral Image Analysis
III.
2D image Analysis
IV.
Spectral Analysis and Chemical Analysis for Ground truth verification
I)
Data Collection
Four different data sets will be considered for addressing the above mentioned study
a) Hyperspectral Images will be
 Purchased from ISRO. Visible and near infra-red (VNIR, or 400 nm to 1000 nm) and
shortwave infra-red (SWIR, 900 nm to 2500 nm) range images will be considered.
 The European Space Agency’s Multispectral Instrument (MSI) Sentinel-2 also
provides (open source) 10-meter resolution, multispectral images every 10 days (2015present) to support vegetation, land cover, and environmental monitoring. Sentinel2 MSI acquires 13 spectral bands ranging from Visible and Near-Infrared (VNIR) to
Shortwave Infrared (SWIR) wavelengths along a 290-km orbital swath.
b) High resolution 2D Satellite Images will be purchased from ISRO.
c) 2D Images acquired by UAV: A drone mounted with multi sensors will be used to capture
multi spectral images from the area of study.
d) Soil samples will be collected from the area of study for chemical and spectral analysis
Table 1: Summary of Advantages and Disadvantages of the data sets considered
Platform
Advantages
Disadvantages
Satellites
 Provides large areas topsoil
 Spectral measurements affected
information
by atmospheric absorptions
 Inaccessible areas information
 Low signal-to-noise ratio
is obtained
 Mixed pixels contain soil surface
 Auxiliary data is present
and vegetation etc
 Temporal resolution is
 Geometric, atmospheric
Consistent
corrections needed
 Periodic Data available (every
10 days)
 Open source data available
Unmanned
 Flexible data capturing can be
 atmospheric, geometric
Aerial
planned according to weather
corrections are must
Vehicle
condition
 Legal permission required for
 High spatial resolution
data collection
 Limited flight duration
 Limited payload
Ref No. : 182019007047 | Page 8 of 33
II)
Hyperspectral Image Analysis
The Hyper spectral image analysis methodology is as follows:
1. RAW input data
2. Radiometric correction
3. Image to surface reflectance for selection of good bands
4. Selection of Region of Interest
5. Spectral unmixing to remove impurities
6. Post processing
Hyperspectral Remote sensing approaches are fast, nondestructive and cover a large spatial
area and help in providing solution for rapid soil assessment. The Hyperspectral image
analysis will be carried out using the above mentioned steps. These images are generally
degraded by cloud cover, aging of sensors, terrain factors etc. Hence radiometric correction is
the first preprocessing required to be done. Image to surface reflectance will be done using an
open source tool ‘MODTRAN 4’ atmospheric model approach. This is done to select desired
good bands. Region of interest is selected, followed by spectral unmixing using non-negatively
constrained least squares (NCLS) algorithm. The digital maps are generated using post
processing steps shown in fig2.
Correlated coefficient can be found between soil organic matter data (spectrum of soil sample) and
other selected variables (image) to compute correlation analysis using the following equation.
𝑟=
′
′
∑𝑁
𝑛=1(𝑅𝑛 − 𝑅 )(𝑆𝑂𝑀𝑛 − 𝑆𝑂𝑀 )
′ 2
′ 2
√∑𝑁
𝑛=1(𝑅𝑛 − 𝑅 ) (𝑆𝑂𝑀𝑛 − 𝑆𝑂𝑀 )
Where r is correlation coefficient, R is selected variables (e.g. spectral reflectance/ color indices etc),
n is the no of soil samples, SOM is soil organic matter, R’ and SM’ is mean value.
Multivariate statistical regression can be used to find relation between variables and soil matter.
Discriminate score can be found using the following linear equation
D= B0 +B1X1 + B2X2 +………..+BnXn
D is Discriminate score, B0, n are estimated constants and coefficients, Xn are variables.
Five color indices brightness index (BI), coloration index (CI), hue index (HI), redness index (RI)
and saturation index (SI) will be selected to analyze the correlation between image features and
soil organic matter content.
Ref No. : 182019007047 | Page 9 of 33
Post Processing: Correlation of Features of Hyper spectral data and data
from soil spectrometer
Image
Chemical / Spectrometer Raw
Data
Calibration
Calibration
Extract Spatial
Signature
Extract Spatial Signature
Spectral
Matching
Hyperspectral image
Fusion of
Parameters
Digital Map
Generation
Fig 2 Post processing for digital map generation
Where
BI is brightness index given by 𝐵𝐼 = √
CI is coloration index given by CI =
HI is hue index given by HI =
𝑅 2 +𝐺 2 +𝐵2
R−G
3
R+G
2∗R−G−B
G−B
𝑅2
RI is redness index given by 𝑅𝐼 = 𝐵∗𝐺3
(𝑅−𝐵)
SI is Saturation Index given by 𝑆𝐼 = (𝑅+𝐵)
R=681nm, G=569nm, b=487nm
Concentration of soil ORGANIC matter is given by
SOM = 14.375 +12.456CI – 8.276HI
Ref No. : 182019007047 | Page 10 of 33
Similarly correlation coefficient will be found for other properties of the soil. Other spectral indices
which will be considered are Normalized Differences Vegetation Index (NDVI), Transformed
Vegetation Index (TVI), Enhanced Vegetation Index (EVI), Soil Adjusted Total Vegetation Index
(SATVI), Soil-Adjusted Vegetation Index (SAVI), Moisture Stress Index (MSI), Green
Normalized Difference Vegetation Index (GNDVI), Green-Red Vegetation Index (GRVI), Land
Surface Water Index (LSWI), Transformed Soil Adjusted Vegetation Index (TSAVI), Modified
Soil Adjusted Vegetation Index (MSAVI), the Second Modified Soil Adjusted Vegetation Index
(MSAVI2), Weighted Difference Vegetation Index (WDVI), the Second Brightness Index (BI2)
and Vegetation (V) [10] to find various signatures of the soil and evaluate the soil property.
III 2D image Analysis
The data from satellite or UAV shall be used for 2d image analysis. General block diagram is
shown in fig 3. Initially pre-processing of data is required to remove noise, atmospheric corrections
etc. Various textural features can be extracted from different bands for feature extraction to analyze
soil properties. A feature database can be generated which is used for training and testing of Neural
Networks (NN). Convolutional neural network (CNN) model can extract contextual information
surrounding an observation to improve the prediction accuracy over conventional methods. NN
is flexible and works well for complex problems with high prediction accuracy. Non-linear local
spatial relationships of neighboring pixels can be found by CNN models from 2 D images. CNN
model can be trained to predict soil organic carbon at various depths. From the input image, CNN
is able to recognize contextual information and extract multi-scale information automatically.
Multilayer perceptron NN (MLP-NN) and Radial Basis Function NN (RBF-NN) can be considered
for the analysis. MLP-NN has three layers, input, hidden, and output. The number of input neuron
is equal to number of input variables, the number of hidden neuron must be computed and the
number of output neuron can be equal to number of outputs. Behavior of NN is determined by
weights between the three layers. These weights are initiated randomly and then, updated using
the back-propagation algorithm through iteration processes. RBF-NN is also a three layer model
but the hidden layer of RBF-NN is referred to the RBF units. K-means algorithm is used to cluster
the input neurons into new space.
Gaussian Process (GP) formulates the regression model where its parameters follow a Gaussian
distribution using a Bayesian statistics. From soil salinity data set D= (Xi, yi), i=1, 2, 3….,m) with
X is the matrix for m input values and y outputs.
𝑛
𝑦̂ = 𝑓(𝑥) = ∑ 𝛼𝑖 𝐾( 𝑋𝑖 , 𝑋)
𝑖=1
Deep learning is used in multispectral image classification for land cover classification. A deep
NN has a number of three-dimensional hidden layers, each convolution or pooling layer learns to
detect different features of the input images. Convolution takes the input images through a set of
convolutional filters each of which detects and enhances certain features from the images. A
pooling operation merges similar features by performing non-linear down-sampling. Pooling
makes the features robust against noise. All the convolutional and pooling layers are finally
“flattened” to the fully connected layer. Deep NN are “data-hungry” as they work better with large
Ref No. : 182019007047 | Page 11 of 33
volumes of data. Data augmentation is to generate new samples by modifying the original data
without changing its meaning. Hence we can rotate it by 90, 180, and 270 degrees.
Satellite/2D image
Preprocessing
Feature extraction
Training Database creation
Deep neural
networks
RBF neural
networks
MLP neural
networks
Support vector
regresion (SVR)
Gaussion process
Advanced machine learning algorithms
Performance assessment, validation and comparison
Generation of digital map of soil profile
Fig 3 Flow chart for 2D image analysis
Support vector regression (SVR) is a regression model of support vector machines, and is
considered one of the most powerful technique in advanced machine learning for detecting soil
organic carbon, soil salinity etc. The advantage of using SVR is that it works well with small
training samples. Nu-SVR can be used here to generate the following regression function
𝑛
𝑓(𝑥) = ∑(𝜆𝑖 − 𝜆∗𝑖 )𝑘(𝑥𝑖 − 𝑥) + 𝑏
𝑖=1
Where 𝜆𝑖 𝑎𝑛𝑑 𝜆∗𝑖 denote Lagrange multipliers and k(xi, x) is the RBF kernel function.
The performance of different models can be assessed and compared using the root-mean-square
error (RMSE), the mean absolute error (MAE), and the correlation coefficient (r).
Ref No. : 182019007047 | Page 12 of 33
RMSE = √∑𝑛𝑖=1
(𝑦̂𝑖 − 𝑦𝑖 )2
𝑛
1
MAE = 𝑛 √∑𝑛𝑖=1|𝑦̂𝑖 − 𝑦𝑖 |
r=
̅̂
̅̅
∑𝑛
̅) (𝑦̂𝑖 −𝑦)
𝑖=1(𝑦𝑖 −𝑦
̅̂
̅̅2
̅)2 (𝑦̂𝑖 −𝑦)
√∑𝑛
𝑖=1(𝑦𝑖 −𝑦
where 𝑦̂,
̅ , 𝑦̅̂ are mean of measured and predicted
𝑖 𝑦𝑖 are computed and measured values. 𝑦
values. Performance assessment, validation and comparison will be done with ground truth
parameters followed by digital map generation of soil profile.
IV Chemical and Spectral Analysis
The soil samples are to be collected at 0–10 cm depth as composite samples over an area of
6×6 m which has to be air-dried, ground and sieved (≤2 mm). This mixture has to be thoroughly
mixed before analyzing. Soil Organic Compound (SOC) can be measured as total oxidized carbon
using wet oxidation. The soil particle size distribution can be obtained by the pipette method.
Concentration of chemical analysis
C, NO3-, CO3, SOC, Ph, EC
Concentration for soil samples
Soil samples
Air drying and
Grinding of samples
Spectral reflectance
of soil samples
Identification of
significant bands
using regression
model
First derivative of
spectral signature
Spectral Analysis
Fig 4 Flow chart for spectral and chemical analysis
Spectral analysis can be done by calculating spectral reflectance across the 350–2500 nm
wavelength range using a field spectroradiometer with a high intensity contact probe. A
spectroradiometer is a special kind of spectrometer that can measure radiant energy (radiance and
irradiance). The Field Specspectroradiometer is specifically designed for field environment remote
Ref No. : 182019007047 | Page 13 of 33
sensing to acquire visible near-infrared (VNIR) and short-wave infrared (SWIR) spectra with high
signal-to-noise ratio and superior repeatability of results for better identification and analysis of
materials.
3.2Time Schedule of activities giving milestones through BAR diagram.
Activity
year
1
2
3
Receive Grant
Place order of
Equipment
Sample Data
collection
Chemical and
spectral analysis
Lab set up
Literature Review
Identify algorithm
& model
Data analysis
Submission of
Report
Paper Publication
3.3Suggested Plan of action for utilization of research outcome expected from the project.
To build capacity of the farmers in the region for improving rural livelihoods through knowledge
sharing and dissemination strategy.
Advanced machine learning models can be used for mapping soil salinity, thus, providing a useful
tool for assisting farmers and the policy maker in choosing better crop types in the context of
climate change.
 Utilization of 9 million hectare (mh) of saline and alkaline soils available in India for fish
culture, a second blue revolution can be achieved in shorter period of time span. Centre for
Applied Research on Problematic Soils (CARPS) will work on applied research in order to
utilize the problematic soils for food production.
 Experts from the different institutions can collaborate and initiate the field based research
in multi locations and develop a package of practice in a short period of time.
Soil salinity detection will help the development departments and farmers to measure the extent
and severity of the problem and take appropriate measures available. This benefits society at large.
Dissimation of knowledge
Ref No. : 182019007047 | Page 14 of 33
The proposed project will be handled with multi-institutional linkage by inviting the industry
professionals, students, policy makers, developmental officers, bankers, NGOs, farmers,
entrepreneurs, corporate, consultants, industrialists and press people, and are expected to suggest
few key ideas. The core research team will take up need based research and reach out the
beneficiaries at a faster rate
Project Location
CARPS, NBSSLUP ,
RNSIT and other
research institutes
Krishi Vigyan Kendras
State level dept. of
Agri/horti/fisheries
Farmers
Early detection of soil problems will help the development departments and farmers to measure
the extent and severity of the problem and take appropriate measures available. By and large, the
project enables environment and soil conservation which benefits society at large.
Research publications on above topics with the progress of completed work.
Shivakumar, M., D. Seenappa, K. Manjappa Venkatappa, A. V. Swamy, 2013. Saline/Alkaline
Soils-New Resources for Aquaculture. Environment & Ecology 31 (1): 135—138, January—
March 2013
3.4Environmental impact assessment and risk analysis.
The experimental design is environmental friendly, where only organic and recommended
inorganic amendments are suggested as remedial measures and it will be taken up in consultation
with specialized subject experts of agriculture, horticulture and fisheries. The proposed remedial
measures will enhance soil fertility and conserve soils.
4. Expertise:
4.1 Expertise available with the investigators in executing the project:
Ref No. : 182019007047 | Page 15 of 33
Dr. Vipula
Singh
Dr. Usha B
S
Dr
Shivakumar
PhD in Image processing, has Analysis of lossy to near-lossless compression
experience in working with of hyperspectral imagery using prediction and
Hyperspectral images.
multistage vector quantisation, Sensing and
Imaging ISSN (online) 1557-2072, ISSN
(print),, 2019
Significance of pre-processing and its impact
on lossless hyperspectral image compression,
The Imaging Science Journal, 2017
Low-Complexity and High-Quality Image
Compression Algorithm for Onboard
Satellite, International Journal of Computer
Applications (IJCA), 2012
PhD in Image processing, Has Modelling of edge detection and segmentation
experience in working with 2 D algorithm for pest control in plants, Image
images and classification and
enhancement, noise removal, classification
segmentation algorithms
algorithms
Professor of Aquatic Biology
and Head of Department of
Aquatic
Environment
Management has worked on the
secondary salinization issue for
more than 15 years and under
sponsorship of RKVY handled
the developmental project
“Reclamation of saline and
alkaline
soils
through
aquaculture” in 258 villages of
Karnataka, India and developed
package of practice for
aquaculture in saline and
alkaline soils.
Shivakumar, M., D. Seenappa,
K. Manjappa Venkatappa, A.
V.
Swamy,
2013.
Saline/Alkaline
Soils-New
Resources
for
Aquaculture. Environment &
Ecology 31 (1): 135—138,
January—March 2013
To take up applied research on the same issue
a centre called “Centre of Applied Research
on Problematic Soils (CARPS)” has been
established at Mangaluru and it is functioning
since 2018.
Ref No. : 182019007047 | Page 16 of 33
4.2 Summary of roles/responsibilities for all Investigators:
Name of the Investigators Roles/Responsibilities
1. Dr. Vipula Singh : Hyperspectral image analysis
2. Dr. Usha B S : 2 D image analysis
3. Dr Shivakumar : Soil analysis and ground truthing
4.3Key publications published by the Investigators pertaining to the theme of the proposal
during the last 5 years
S.N Author(s)
Title
Name of Journal Volu
Page
Year
o.
me
Vipula Singh Decluttering using
Sensing and
2019
1
Vol.
1-21
Smitha N,
wavelet based higher
Imaging ISSN
20:2
2
Vipula Singh
A. S.
Mamatha
3
Vipula Singh
A. S.
Mamatha &
Rajath
Kumar M P
Vipula Singh
Sujatha S.
Swamy;
Mamatha
A.S;
4
order statistics and
target detection of GPR
images”,
(online) 15572072, ISSN
(print),
Analysis of lossy to
near-lossless
compression of
hyperspectral imagery
using prediction and
multistage vector
quantisation
Significance of preprocessing and its
impact on lossless
hyperspectral image
compression
Low-Complexity and
High-Quality Image
Compression
Algorithm for Onboard
Satellite
The Imaging
Science Journal
Vol.65
/3
180190
April
2017
The Imaging
Science Journal
Vol.65
/5
270281
May 2017
International
Journal of
Computer
Applications
(IJCA)
5
Usha B S,
Samit Desai
Medical image
transcoder for
telemedicine based on
wireless
communication devices
IEEE Xplore
6
Usha B S, KS
Srinivas, , S
Sandya, SR
Rupanagudi
Modelling of edge
detection and
segmentation algorithm
for pest control in
plants
International
Conference in
Emerging Trends
in Engineering,
7
Usha B S,
Sandya S,
Sreedevi T
Papsmear Image
based Detection of
Cervical Cancer
International
Journal of
Computer
Applications
(IJCA)
June 2012
DOI: 10
.1109/I
CECTE
CH.201
1.59416
29
293-295
Vol
45–
No.20
July 2011
July 2011
(0975 – May
8887)
2012
Ref No. : 182019007047 | Page 17 of 33
8
Usha B S,
Sandya S,
Shruthi G
9
Usha B S,
Sandya S,
A Novel Approach
for Speckle
Reduction and
Enhancement of
Ultrasound Images
Size and ShapeBased Ovarian
Abnormality
Detection of
Ultrasound Images
International
Journal of
Computer
Applications
(IJCA)
Journal of
1. Vol 1
,No 4
Biomedical
Engineering
and Medical
Imaging
10
Usha B S,
Sandya S,
Abnormality
1.
Detection in Ovarian
Ultrasound Images
using Active
Contours
11
Usha B S,
Sandya S,
12
Usha B S,
Sandya S,
Ultrasound Ovary
Image Classification
Using Kσ-Classifier
Measurement of
ovarian size and
shape parameters
Shivakumar,
M,
Naveenkum
ar, B.T.,
Shivananda
Murthy,
Ramachandr
a Naik,
A.T., and C.
Vasudevapp
a
Shivakumar,
M., D.
Seenappa,
K.
Manjappa
Venkatappa,
Vol
45–
No.20
(0975 – May
8887)
2012

DOI:

10.100
7/9788132211570_31
Sep 2014
(2014)
198202
2013
International
Journal of
Science,
Environment
and Technology
10.110
9/IND
CON.2
013.67
26079
Vol. 1,
No 5,
2012,
491 –
498.
Saline/Alkaline Soils- Environment &
New Resources for
Ecology
Aquaculture
31 (1):
135—
138
January
—March
2013
Studies on Effect of
Claw-Ablation on
Growth and Survival
of Macrobrachium
Rosenbergii (De
Man).
2013 Annual
IEEE India
Conference
(INDICON)
2015
2012
Ref No. : 182019007047 | Page 18 of 33
A. V.
Swamy
Shivakumar
M
Organic Farming:
Complex RealitiesAn opinion paper
Modern
Concepts and
Developments
in Agronomy,
Vol.2,
Issue 1.
2018
4.4Bibliography
[1] Zhuo Luoa, Liu Yaolin, Wu Jiana, Wang Jing, “Quantitative Mapping Of Soil Organic
Material Using Field Spectrometer And Hyperspectral Remote Sensing”, The International
Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol.
XXXVII. Part B8. Beijing 2008
[2] Andreas Kamilaris, Francesc X. Prenafeta-Boldú, “Deep Learning in Agriculture: A Survey”,
Institute for Food and Agricultural Research and Technology (IRTA)
[3] W. Dean Hively, GregoryW.McCarty, James B. Reeves III, MeganW. Lang, Robert A.
Oesterling, and Stephen R. Delwiche, “Use of Airborne Hyperspectral Imagery to Map Soil
Properties in Tilled Agricultural Fields”, Hindawi Publishing Corporation, Applied and
Environmental Soil Science Volume 2011, Article ID 358193, 13 pages
doi:10.1155/2011/358193
[4] Michael L. Whiting, et al, “Hyperspectral mapping of crop and soils for precision agriculture
Remote Sensing and Modeling of Ecosystems for Sustainability”, Proc. of SPIE Vol. 6298,
62980B, (2006) · 0277-786X/06/$15 · doi: 10.1117/12.681289
[5] Daniel Vogt, Darlene Zabowski, L. Monika Moskal, “Hyperspectral Analysis of Soil Nitrogen,
Carbon, Carbonate and Organic Matter Using Regression Trees”, Sensors 2012, 12, 1063910658; doi:10.3390/s120810639.
[6] Eliza E Camaso et al, “Landcover mapping using LIDAR data and aerial image and soil fertility
degradation assessment for rice production area in Quezon, Nueva Ecija, Philippines”,
International Journal of computer and system engineering Vol 11,No 7 2017
[7] José Padarian, Budiman Minasny, Alex B. McBratney, “Using deep learning for digital soil
mapping”, Published by Copernicus Publications on behalf of the European Geosciences
Union, 2019
[8] Pham Viet Hoa et al, “Soil Salinity Mapping Using SAR Sentinel-1 Data and Advanced
Machine Learning Algorithms: A Case Study at Ben Tre Province of the Mekong River Delta
(Vietnam)”, Remote Sens. 2019, 11, 128;
[9] Michael Vohland et al , “Quantification of Soil Properties with Hyperspectral Data:
Selecting Spectral Variables with Different Methods to Improve Accuracies and Analyze
Prediction Mechanisms”, Remote Sens. 2017, 9, 1103; doi:10.3390/rs9111103
[10] A. Gholizadeh, D. Žižala, M. Saberioon, and L. Borůvka, “Soil organic carbon and texture
retrieving and mapping using proximal, airborne and Sentinel-2 spectral imaging,” Remote
Sens. Environ., vol. 218, no. December, pp. 89–103, 2018.
[11] Soil Contamination Mapping with Hyperspectral Imagery: Pre-Dnieper Chemical Plant
(Ukraine) Case Study Sergey A. Stankevich, Mykola M. Kharytonov, Anna A. Kozlova,
Vadym Yu. Korovin, Mykhailo O. Svidenyuk and Alexander M. Valyaev
Ref No. : 182019007047 | Page 19 of 33
[12] B. S. Das, M. C. Sarathjith, P. Santra, R. N. Sahoo, R. Srivastava, Routra, S. S. Ray,
“Hyperspectral remote sensing: opportunities, status and challenges for rapid soil assessment
in India”, Current Science, Vol. 108, No. 860 5, 10 March 2015
[13] P. Nila Rekha, R. Gangadharan, S.M.Pillai, G.Ramanathan, A. Panigrahi, “Hyperspectral
image processing to detect the soil salinity in coastal watershed”, ICOAC 2012
[14] K.N. Chaudhari, Rojalin Tripathy, N.K. Patel, “Spatial wheat yield prediction using crop
simulation model”, GIS, remote sensing and ground observed data, Journal of
Agrometeorology 12 (2): 174-180 (Dec 2010)
[15] R. R. Ajai, a. S. Arya, P. S. Dhinwa, S. K. Pathan, and K. Ganesh Raj, “Desertification/land
degradation status mapping of India,” Curr. Sci., vol. 97, no. 10, pp. 1478–1483, 2009.
[16] Indian Space Research Organization, “Desertification and Land Degradation - Atlas of
India,” p. 252, 2013.
[17] M. a. E. AbdelRahman, a. Natarajan, R. Hegde, and S. S. Prakash, “Assessment of land
degradation using comprehensive geostatistical approach and remote sensing data in GISmodel builder,” Egypt. J. Remote Sens. Sp. Sci., no. xxxx, 2018.
[18] M. a. E. Abdelrahman, a. Natarajan, C. a. Srinivasamurthy, and R. Hegde, “Estimating soil
fertility status in physically degraded land using GIS and remote sensing techniques in
Chamarajanagar district, Karnataka, India,” Egypt. J. Remote Sens. Sp. Sci., vol. 19, no. 1, pp.
95–108, 2016.
5. List of Projects submitted/implemented by the Investigators
5.1 Details of Projects submitted to various funding agencies:
Month Role
of
as
S. Title
Cost in
submiss PI/C
No
Lakh
ion
o- PI
1
Indigenous method using
49L
PI
electromagnetic ground
6/10/20
penetration radar for detection of
18
Soil Salinity and suitability to aid
and Guide Farmers of Karnataka
2
3
4
Lossless Hyperspectral Image
Compression using Preprocessing
and prediction
Hyperspectral Image Analysis For
Oil Explorations
Concept proof, through prototype
building of indigenous
electromagnetic ground
penetration radar for detection of
14.5L
16 L
Agency
Status
VGST
Reject
CESEM
2017
PI
ISRO
Reject
Oct
2016
PI
SERB
Reject
July
2014
Co
PI
VGST
Reject
CESEM
Ref No. : 182019007047 | Page 20 of 33
unexploded ordinance and anti
personnel non-ferrous mines
5
Hyperspectral image analysis for
oil explorations
14.47 L
Oct
2016
5.2Details of Projects under implementation Nil
S. No Title
Cost in Lakh
Duration
PI
SERB
Role as PI/Co-PI
Reject
Agency
5.3Details of Projects completed during the last 5 years
S. No
Title
Cost in
Lakh
Duration
Agency
2014-2016
Role
as PI/CoPI
PI
1
Detection of Anti-personnel
Mines using Indigenous
Ground Penetrating Radar
(GPR)
20,17,000/-
2
Analysis and implementation
of Kirchoff’s migration and
Omega-K algorithms for
focusing Ground Penetrating
Radar Image
9,75,000/-
2014-2016
PI
LRDE
(CARS*02)
AICTE
(RPS*)
6. List of facilities being extended by parent institution(s) for the project implementation.
6.1 Infrastructural Facilities
Sr. No.
Infrastructural Facility
1.
2.
3.
4.
5.
6.
7.
8.
Workshop Facility
Water & Electricity
Laboratory Space/ Furniture
Power Generator
AC Room or AC
Telecommunication including e-mail & fax
Transportation
Administrative/ Secretarial support
Yes/No/ Not required Full
or sharing basis
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Ref No. : 182019007047 | Page 21 of 33
9.
10.
11.
12.
Information facilities like Internet/Library
Computational facilities
Animal/Glass House
Any other special facility being provided
Yes
Yes
Not required
Not required
6.2Equipment available with the Institute/ Group/ Department/Other Institutes for the
project:
Generic Name of
Equipment
Model, Make & year
of purchase
Remarks including
accessories available
and current usage of
equipment
General Purpose
Processors- 2 No.s
Dell precision T1700
BN8R6C2, 2016
Ground Penetrating
Radar Lab, project
Lab
Equipment
available with
PI & group
PI's Department
Dell Optiplex 9010
G6MQ6C2, 2016
Matlab 12 with Simulink,
DSP system tool Box,
Signal Processing Tool
Box, Image processing
tool Box etc.
Other Institute(s) Field Spectrometer
in the region
2012
It is used for teaching
Lab to UG students
and also for project
implementation
Spectral analysis of
soil samples for
research work
7. Name and address of experts/ institution interested in the subject / outcome of the project.
1. Dr Shivakumar M, Professor and Head, CARPS, Fisheries College Campus Mangalore
575002, Karnataka, India
2. Dr. S. Dharumrajan, Senior Scientist (Soil Science) Regional Centre, NBSS&LUP, Bangalore
Ref No. : 182019007047 | Page 22 of 33
Budget Details
Institution wise Budget Breakup :
Budget Head
RNS Institute of Technology
Manpower
Total
1,116,000
1,116,000
90,000
90,000
Travel
150,000
150,000
Equipment
500,000
500,000
50,000
50,000
Other cost
325,000
325,000
Overhead
0
0
2,231,000
2,231,000
Consumables
Contingencies
Total
Institute Name :
RNS Institute of Technology
Year Wise Budget Summary
(Amount in INR) :
Budget Head
Year-1
Manpower
Year-2
Year-3
Total Amount
372,000
372,000
372,000
1,116,000
Consumables
30,000
30,000
30,000
90,000
Travel
50,000
50,000
50,000
150,000
500,000
0
0
500,000
Contingencies
15,000
15,000
20,000
50,000
Other cost
75,000
150,000
100,000
325,000
Overhead
0
0
0
0
1,042,000
617,000
572,000
2,231,000
Equipments
Grand Total
Manpower Budget Detail (Amount in INR) :
Designation
Justifiocation
Data Collection, Execution and Report
Generation
Research Assistant
Year-1
Year-2
372,000
Year-3
372,000
Total Amount
372,000
1,116,000
Consumable Cost Detail (Amount in INR) :
Justification
Books, Journals, Printouts, Stationary items etc
Year-1
Year-2
30,000
Year-3
30,000
Total Amount
30,000
90,000
Travel Cost Detail (Amount in INR) :
Justification (Inland Travel)
Travel for Sample Collection and Paper presentation
Year-1
Year-2
50,000
Year-3
50,000
Total Amount
50,000
150,000
Equipment Cost Detail (Amount in INR) :
Generic Name
Model No.
(Make)
Drone
Agrivison
(IO Techworld
Avigation Pvt.
Ltd.)
Desktop
Computer
(GPU)
Dell XPS 8930
(Dell)
Justification
Quantity
Spare time Estimated Cost (INR)
To capture multi spectral images of soil
1
50 %
GPU is needed for the processing of hyper spectral
and satellite images.
1
5%
3,00,000
2,00,000
Contingency Cost Detail (Amount in INR) :
Justification
For uncertainty in data collection for field trials.
Year-1
Year-2
15,000
Year-3
15,000
Total Amount
20,000
50,000
Overhead Detail (Amount in INR) :
Justification
Year-1
Nil
Year-2
0
Other Budget Detail
Description
Publication Charges
Purchase of data set
Year-3
0
Total Amount
0
0
(Amount in INR) :
Justification
Paper publications in Journals and Conferences.
Hyper spectral and satellite images need to be
purchased from ISRO or other agencies.
Soil sample collection Soil samples of the selected region has to be
during field trials
collected physically every 100 meters to verify
the ground truth.
Chemical and Spectral analysis of soil samples
Testing of soil
need to be done from the research labs. per
samples
sample cost need to be paid to the organisation
for testing.
Year-1
Year-2
Year-3
Total Amount
0
25,000
25,000
50,000
50,000
50,000
0
100,000
25,000
25,000
25,000
75,000
0
50,000
50,000
100,000
Ref No. : 182019007047 | Page 23 of 33
PROFORMA FOR BIO-DATA (to be uploaded)
1. Name and full correspondence address
Dr. Vipula Singh,
HoD, Dept. of ECE,
RNS Institute of Technology, Bengaluru-560098
2. Email(s) and contact number(s) vipulasingh@yahoo.com, 8105917500
3. Institution RNSIT
4. Date of Birth 10/09/1971
5. Gender (M/F/T) F
6. Category Gen/SC/ST/OBC GEN
7. Whether differently abled (Yes/No)
8. Academic Qualification (Undergraduate Onwards)
Degree
BE
Year
1992
Mtech
2003
PhD
2009
Subject
Electronics
Engineering
Electronics
Engineering
Development of
Image
Compression
Algorithms
using Soft
Computing
Approaches
University/Institution % of marks
78.34%
MNIT Bhopal
(REC.)
81.57%
VNIT (REC)
Nagpur
Guru Gobind Singh
Indra Prastha
University Delhi
Ph.D thesis title, Guide’s Name, Institute/Organization/University, Year of Award.
Development of Image Compression Algorithms using Soft Computing Approaches, Dr.
Navin Rajpal, Guru Gobind Singh Indra Prastha University, 2009
10. Work experience (in chronological order).
S.No.
Positions
held
1
Design
engineer
Name of
From
the
Institute
Punjab
Jan 1993
Communica
tions ltd.
To
Pay Scale
Dec 1994
6000
consolidated
Ref No. : 182019007047 | Page 24 of 33
Lecturer
Lecturer
Lecturer
Lecturer
Lecturer
Assistant
Professor
(Govt
Undertakin
g) Mohali
Madhav
Instiyute of
Technology
and
Science,
Gwalior,
Madya
Pradesh
Institute of
Engineering
and
Technology
, Indore,
Madhya
Pradesh
Communica
tions Sri
Ram Dev
Baba
Kamla
Nehru
Engineering
College,
Nagpur,
Maharashtr
a
Institute of
Integral
Technology
, Lucknow
Uttar
Pradesh
Appejay
College of
Eng.
Gurgaon
Haryana
PES
Institute of
Technology
(PESIT)
Bangalore
June 1995
June 1998
8000-27513500
Aug 1998
April 2000
8000-27513500
Aug 2000
March
2001
8000-27513500
April 2002
April 2003
8000-27513500
May 2003
July 2004
8000-27513500
Aug 2004
October
2008
12000-42018300
Ref No. : 182019007047 | Page 25 of 33
Associate
Professor
Professor &
Head of the
department
Professor &
Head
VNRVJIET Nov 2008
Hyderabad
Geethanjali June 2010
College of
Engineering
&
Technology
Hyderabad
RNSIT
April 2011
Bangalore
May 2010
March
2011
Till date
12000-42018300
21450-67528875
20900-50022400
11. Professional Recognition/ Award/ Prize/ Certificate, Fellowship received by the applicant.
S.No
1
2
3
4
Name of Award
National Scholarship from 5th to 12th
standard
State merit scholarship from 5th to
12th standard
Bronze Medal, State fellowship for
12th standard
Gold Medal in M.Tech
Awarding Agency
Govt. of India
Year
1981-1988
Govt. of Himachal
Pradesh
Govt. of Himachal
Pradesh
NIT, Nagpur
1981-1988
1988
2003
12. Publications (List of papers published in SCI Journals, in year wise descending order).
S.N
o.
Author(s)
Title
Smitha N,
Vipula
Singh
Decluttering using
wavelet based higher
order statistics and
target detection of
GPR images”,
A. S.
Mamatha &
Vipula
Singh
Analysis of lossy to
near-lossless
compression of
hyperspectral
imagery using
prediction and
multistage vector
quantisation
Significance of preprocessing and its
impact on lossless
hyperspectral image
compression
A. S.
Mamatha ,
Vipula
Singh &
Name of
Journal
Sensing and
Imaging ISSN
(online) 15572072, ISSN
(print),
Volum Page
e
Vol.
1-21
20:2
Year
The Imaging
Science
Journal
Vol.65/ 180-190
3
April
2017
The Imaging
Science
Journal
Vol.65/ 270-281
5
May
2017
2019
Ref No. : 182019007047 | Page 26 of 33
Rajath
Kumar M P
N. Smitha,
D. R. Ullas
Bharadwaj,
S. Abilash,
S. N.
Sridhara,
Vipula
Singh
N. Smitha,
Vipula
Singh
SujithKuma
r S B,
Vipula
Singh
Sujatha S.
Swamy;
Mamatha
A.S; Vipula
Singh
Vipula
Singh
Vipula
Singh
Vipula
Singh
Vipula
Singh
Effect of Ground
Undulation and
Variation in velocity
of vehicle on
Migration of SFCW
GPR data for
landmine detection
Exploration
Geophysics
Journal
(CSIRO)
, Dec
2016
Kirchhoff migration
to focus Ground
Penetrating Radar
Images for Landmine
detection
Automatic Detection
of Diabetic
Retinopathy in Nondilated RGB Retinal
Fundus Images
Low-Complexity and
High-Quality Image
Compression
Algorithm for
Onboard Satellite
Patents on
Image/video
Compression
International
journal of Geo
Engineering
2016
International
Journal of
Computer
Applications
(IJCA)
International
Journal of
Computer
Applications
(IJCA)
Recent
Patents on
Signal
Processing
Digital
Journal of
Watermarking: A
Selected
Tutorial
Areas in
Telecommuni
cations
(JSAT)
Recent Patents on
International
Image Compression – Journal on
A Survey
Recent
Patents on
Signal
Processing
A Neuro-Wavelet
International
model using Fuzzy
Journal of
vector Quantization
Image and
for efficient Image
Graphics
Compression
(IJIG)
Vol.47
/17
,June
2012
June
2012
vol 1
101-115
2011
11-21
January
E, 2011
vol 2
47-62
2010
vol 9
no 2
299-320
2009,
Ref No. : 182019007047 | Page 27 of 33

Book chapter in” (Mathematics Research Developments)
13. Detail of patents.
14. Books/Reports/Chapters/General articles etc.
S.No
Title
Author’s Name
Publisher
1
Development of Image
Compression Algorithms
using soft computing
approaches
Image Processing with
MATLAB & LabVIEW
Wavelets: Classification,
Theory & Applications
Vipula Singh
Lambert
Academic
publishing
Germany
Elsevier
2012
Publisher
Nova publishers
2
3
Vipula Singh
Vipula Singh
Year of
Publication
2011
15. Any other Information (maximum 500 words)
Ref No. : 182019007047 | Page 28 of 33
PROFORMA FOR BIO-DATA (to be uploaded)
1. Name and full correspondence address
Dr. Usha B S
Assistant Professor, Dept. of ECE,
RNS Institute of Technology, Bengaluru-560098
2. Email(s) and contact number(s) bsusha@gmail.com, 9663075706
3. Institution RNS Institute of Technology, Bengaluru-98
4. Date of Birth 11/03/1976
5. Gender (M/F/T) F
6. Category Gen/SC/ST/OBC GEN
7. Whether differently abled (Yes/No) No
8. Academic Qualification (Undergraduate Onwards)
Degree
BE
Year
1997
Mtech
2003
PhD
2017
Subject
Electronics and
Communication
Engineering
Electronics
Engineering
Ovarian
abnormality
detection using
geometrical and
statistical
features
University/Institution % of marks
Mysore University, 62%
SJCE, Mysore
VTU, SJCE,
Mysore
VTU, Belgaum
76.5 %
9. Ph.D thesis title, Guide’s Name, Institute/Organization/University, Year of Award.
“Ovarian abnormality detection using geometrical and statistical features”, Dr. Sandya S,
RNS Institute of Technology, VTU, 2018
Ref No. : 182019007047 | Page 29 of 33
10. Work experience (in chronological order).
S.No.
1
Positions
held
Lecturer
2
Lecturer
3
Lecturer
4
Sr. Lecturer
5
Assistant
Professor
Name of the
Institute
Dr. Ambedkar
Institute of
Technology,
Bengaluru
RV College
of
Engineering,
Bengaluru
RNS Institute
of
Technology,
Bengaluru
RNS Institute
of
Technology,
Bengaluru
RNS Institute
of
Technology,
Bengaluru
From
To
Pay Scale
March 2000
Sep 2002
Rs. 8000
consolidated
March 2003
Oct 2003
8000-27513500
Oct 2003
April 2007
8000-27513500
April 2007
April 2009
10000-32515200
April 2009
Till date
15600-64839100
11. Professional Recognition/ Award/ Prize/ Certificate, Fellowship received by the applicant.
S.No
1
2
Name of Award
President Guide Award
3rd Rank in M.Tech
Awarding Agency
Govt. of India
VTU, Belgaum
Year
1991
2003
12. Publications (List of papers published in SCI Journals, in year wise descending order).
S.N
o.
Author(s)
Title
Usha B S,
Sandya S,
Ultrasound Ovary
Image
Classification
Using KσClassifier
Name of
Journal
In: Toi V.,
Lien Phuong
T. (eds) 5th
International
Conference on
Biomedical
Engineering in
Vietnam.
Volume
Page
Year
IFMBE 198-202
Proceed
ings,
vol 46.
Springe
r, Cham
2015
Ref No. : 182019007047 | Page 30 of 33
Usha B S,
Sandya S,
Abnormality
1.
Detection in
Ovarian Ultrasound
Images using
Active Contours
Journal of
1. Vol 1
,No 4
Biomedical
Engineering
and Medical
Imaging
Usha B S,
Sandya S,
Size and ShapeBased Ovarian
Abnormality
Detection of
Ultrasound Images
ICERECT-12,
LNEE,
Springer,
Usha B S,
Sandya S,
Measurement of
ovarian size and
shape parameters
Usha B S,
Sandya S,
Sreedevi T
Papsmear Image
based Detection of
Cervical Cancer
2013 Annual
IEEE India
Conference
(INDICON)
International
Journal of
Computer
Applications
(IJCA)
International
Journal of
Computer
Applications
(IJCA)
IEEE Xplore
Usha B S,
Sandya S,
Shruthi G
A Novel Approach
for Speckle
Reduction and
Enhancement of
Ultrasound Images
Usha B S,
Medical image
Samit Desai transcoder for
telemedicine based
on wireless
communication
devices
Usha B S,
Modelling of edge
KS Srinivas, detection and
, S Sandya,
segmentation
SR
algorithm for pest
Rupanagudi control in plants
International
Conference in
Emerging
Trends in
Engineering,
13. Detail of patents.
14. Books/Reports/Chapters/General articles etc.
S.No Title
Author’s Name
Publisher
2014


DOI:
Sep 2014
10.1007/
978-8132211570_31
Vol 45–
No.20
10.1109/ 2013
INDCO
N.2013.
6726079
(0975 – May
8887)
2012
Vol 45–
No.20
(0975 –
8887)
May
2012
DOI: 10 July
.1109/IC 2011
ECTEC
H.2011.
5941629
293-295
July
2011
Year of Publication
15. Any other Information (maximum 500 words)
Ref No. : 182019007047 | Page 31 of 33
Ref No. : 182019007047 | Page 32 of 33
Ref No. : 182019007047 | Page 33 of 33
Download