International Journal of Electrical, Electronics and Computer Systems, (IJEECS)
_______________________________________________________________________
GRADUAL TRANSITION DETECTION IN MOVIE VIDEOS: A
CHALLENGE FOR SHOT BOUNDARY DETECTION
ALGORITHM
1
Shraddha C. Nistane, 2Krishna K. Warhade
1
M.E. student at Department of Electronics & Telecommunication, MITCOE, Pune
2
Professor at Department of Electronics & Telecommunication, MITCOE, Pune
Email : 1shraddhanistane@gmail.com, 2krishna.warhade@mitcoe.edu.in
of frames shot uninterruptedly by one camera. There are
several film transitions usually used in film editing to
juxtapose adjacent shots; In the context of shot transition
detection they are usually group into two types i.e abrupt
transition and gradual transition.
ABSTRACT: Automatic shot boundary detection has been
an active research area for nearly a decade and has led to
high performance detection algorithms for hard cuts, fades
and wipes. But finding gradual transition is major
challenge in the presence of camera and object motion. In
this paper, a review of different gradual transition
detection methods is presented. Specially, the review
focuses on dissolve detection in the presence of camera and
object motion.
Abrupt transition (AT) is a sudden transition from one
shot to another, i . e. one frame belongs to the first shot,
the next frame belongs to the second shot. They are also
known as hard cuts or simply cuts [2]. Fig.1 shows
consecutive frames with abrupt transition from star war
movie. Gradual transition (GT) is a transition in which
the two shots are combined using
Index term: Shot boundary detection, gradual transition,
dissolve detection.
I. INTRODUCTION
The increased availability and usage of on-line digital
video has created a need for automated video content
analysis techniques. Most research on video content
involves automatically detecting the boundaries between
camera shots. Shot transition detection is used to split up
a film into basic temporal units called shots; a shot is a
series of interrelated consecutive pictures taken
contiguously by a single camera and representing a
continuous action in time and space [1]. This operation
is of great use in software for post-production of videos.
It is also a fundamental step of automated indexing and
content-based video retrieval or summarization
applications which provide an efficient access to huge
video archives, e.g. user may choose a representative
picture from each scene to create a visual overview of
the whole film and, by processing such indexes, a search
engine can process search items. A digital video consists
of frames that are presented to the viewer's eye in rapid
succession to create the impression of movement. Each
frame within a digital video can be uniquely identified
by its frame index, a serial number. A shot is a sequence
Fig.1. consecutive frames with abrupt transition from
star war I Movie
__________________________________________________________________________
ISSN (Online): 2347-2820, Volume -1, Issue-2, 2013
1
International Journal of Electrical, Electronics and Computer Systems, (IJEECS)
_______________________________________________________________________
Several reviews on shot boundary detection have been
published in the last decade.
III. CONTENT BASED VIDEO INDEXING
AND RETRIEVAL
There are four main processes involved in content-based
video indexing and retrieval [3-5]: video content
analysis, video structure parsing, summarization or
abstraction, and indexing. Each process poses many
challenges. We briefly review and discuss these
challenging research issues as mentioned below.
A.. Video content analysis
Video content analysis is the capability of automatically
analyzing video to detect and determine temporal events
not only based on a single image but also on the basis of
text, audio, speed. Video content analysis is used in a
wide range of areas including entertainment, health-care,
retail, automotive, transport, home automation, safety,
security, network, multimedia, Internet communication,
Mobile communication, Distance education, Sports and
News. Many different functionalities can be
implemented in video content analysis. video motion
detection is one of the simpler forms where motion is
detected with regard to a fixed background scene. More
advanced functionalities include video tracking, object
detection, motion detection, face recognition.
Fig.2. consecutive frames of dissolve transitions
chromatic, spatial or spatial-chromatic effects where one
shot is gradually replace by another. These are also
often known as soft transitions and these shot can be of
various types, e.g., wipes, dissolves, fades [2]. Fig.2
shows consecutive frames of dissolve transition.
Although cut detection appears to be a simple task for a
human being, it is a non-trivial task for computers. Cut
detection would be a trivial problem if each frame of a
video was enriched with additional information
about when and by which camera it was taken. Possibly
no algorithm for cut detection will ever be able to detect
all cuts with certainty, unless it is provided with
powerful artificial intelligence.
B. Video structure parsing
Digital video needs to be properly processed before it c
inserted into a video server. These tasks include
compressing, parsing and indexing a video sequence.
Video parsing is the process of detecting scene changes
or the boundaries between camera shots in a video
stream. The video parsing is a similar process like text
document parsing, but it requires higher level of content
analysis on the basis of pixel, color, edges, motion,
objects etc. Shot boundary detection algorithm process
visual information contained in video frames and can
segment the video into frames with similar visual
information.
While most algorithms achieve good results with hard
cuts, many fail with recognizing soft cuts. Hard cuts
usually go together with sudden and extensive changes
in the visual content, while soft cuts feature shows slow
and gradual changes. A human being can compensate
this lack of visual diversity with understanding the
meaning of a scene. While a computer assumes a black
line wiping a shot away to be "just another regular
object moving slowly through the on-going scene", a
person understands that the scene ends and is replaced
by a black screen.
ARTICLE IN PRESS
II. CHALLENGES IN SHOT BOUNDARY
DETECTION
C. Video summarization
Due to the rapid progress in technology of network and
multimedia, large number of videos is available on the
internet. Video summarization plays an important role in
this context. It helps in efficient storage, quick browsing,
and retrieval of large collection of video data without
losing important aspects. This process is similar to
extraction of keywords or summaries in text document
processing. That is, we need to extract a subset of video
data from the original video such as key frames or
highlights as entries for shots, scenes, or stories.
Combining the structure information extracted from
video parsing and the key frames extracted in video
abstraction, we can build a visual table of contents for a
video.
The major challenge in video segmentation is a
detection of gradual transition in the presence of motion.
A gradual shot transition occurs when the change takes
place over a sequence of frames. The most common
gradual transitions are dissolves and fades (in-out). A
dissolve in a video sequence is a shot transition with the
first shot gradually disappearing while the second shot
gradually appears. A fade of a video sequence is a shot
transition with the first shot gradually disappearing (fade
out) before the second shot gradually appears (fade in).
During the fading transition, two shots are spatially and
temporally well separated by some monochrome frames.
__________________________________________________________________________
ISSN (Online): 2347-2820, Volume -1, Issue-2, 2013
2
International Journal of Electrical, Electronics and Computer Systems, (IJEECS)
_______________________________________________________________________
D. Video indexing
the precision and recall rates can get in a satisfactory
level.
The structural and contents found in content analysis,
video parsing and video summarization are referred as
metadata. So the, Efficient and effective handling
of video documents depends on the availability of
indexes.
Manual indexing is
not
feasible
for
large video collections. Based on the metadata, we can
build video indices and the table of content. After
indexing, these metadata is proceeding for clustering
process where sequence or shots classifies into different
visual categories or an indexing structure. As in many
other information systems, we need schemes and tools to
use the indices and content metadata to query, search,
and browse large video databases. Researchers have
developed numerous schemes and tools for video
indexing and query. However, robust and effective tools
tested by thorough experimental evaluation with large
data sets are still lacking. Therefore, in the majority of
cases, retrieving or searching video databases by
keywords or phrases will be the mode of operation. In
some cases, we can retrieve performance by content
similarity defined by low-level visual features of, for
instance, key frames, and example-based queries.
Yang Xu et al. have proposed 3- DWT based motion
suppression for video shot boundary detection [6]. In
this method adaptive threshold is selected, so that
gradual transition and motion are adequately
discriminated. Under the framework of 3-DWT
framework dramatic motion are characterized. In the
proposed method, motion intensity is extracted and
motion suppression value (MSV) is defined, which is
integrated into histogram based on edges based methods
for video shot boundary detection. This method is more
efficient to detect the gradual transition and also solves
the problem of motion information which suffers from
noise and illumination.
Detection of gradual transition in video sequences using
B-spline interpolation has been proposed by Jeho Nam
and Ahmed H. Tewfik [7]. In this proposed method, the
focus is on to recover the original transition behavior of
edit effect though it was distorted by motion and other
pre-processing operations. In this technique, existing
abrupt shot boundaries are detected. There is undefined
framework for dealing with all gradual transition. To
overcome this problem, B-spline interpolation curve
fitting is used for estimating silent production
IV. RELATED WORK
In this Section, we describe various algorithm in the area
of video segmentation specially gradual transition
detection.
Jun Li et al. have proposed DWT-based shot boundary
detection using support vector machine [8]. In this
technique, shot boundary detection algorithms extract
the color and the edge in different direction from
wavelet transition coefficients. Then a multi-class
support vector machine (SVM) classifier is used to
classify the video shot into three categories: cut
transition (AT), gradual transition (GT) and normal
sequences (NF). To enhance the robustness of the
algorithm, the feature vector from all frames within a
temporal window is formed. This technique is capable
for numerical experiments using a variety of detecting
and discriminating shot transitions in videos with
different characteristics.
The video shot boundary detection using Eigen value,
decomposition and Gaussian transition detection has
been proposed by Ali Amiri [1]. In this method, using
generalized eigen value decomposition, novel shot
boundary algorithm is designed. After comparing the
video frame, the distance function is calculated. This
distance function gives abrupt changes in hard cuts and
semi Gaussian behavior in gradual transition. From this
distance function transition is detected. This approach
significantly increases the effectiveness of shot
boundary task, while at the same time reduces the
computation cost. This approach provides the good
result in recall with a range of 92.4% - 97.2%. But the
accuracy is not satisfactory because it is unable to
predict fast camera motion and flashlight changes.
Vasileios Chasanis et al. have proposed simultaneous
detection of abrupt cuts and dissolves in videos using
vector machines [9]. In the proposed methodology
commonly color histogram and χ2 value features are
used. Here, normalized RGB histogram is chosen. For
each frame normalized histogram is computed with 256
bins for each one of the RGB component as HR, HG and
HB respectively. These three histograms are
concatenated into a 768 dimension vector which
represented final histogram for each frame. Variations of
χ 2 can increases the difference between two histogram
and dissimilarity is obtained by using inter frame
distance. For detecting transition, distance should be less
than the minimum length. The classification of features
is done on support vector machine. It finds an optimal
hyper- plane which separates two points of two classes.
The features which are extracted gives input to SVM
classifier which categorizes transition of the video
Y.-N. Li et al. have proposed a fast shot detection
framework employing pre-processing techniques
including thresholding and bisection-based comparisons
to eliminate non-boundary regions [2]. The factors that
lead to high detection speed in the proposed framework
are three folds. Firstly, a large number of non-boundary
frames are discarded before hard cut and GT detections.
Secondly, the types of shot boundaries can be predicted
in preprocessing stage, so that it can detect the preserved
segments directly using the right detector. Thirdly, the
burden of GT detection is eased, because the duration of
each possible GT is available in advance and it is not
necessary to calculate frame distances for multiple
times. On Simulations and comparisons, significant
speed up is achieved in the proposed framework, while
__________________________________________________________________________
ISSN (Online): 2347-2820, Volume -1, Issue-2, 2013
3
International Journal of Electrical, Electronics and Computer Systems, (IJEECS)
_______________________________________________________________________
sequence into normal transition, abrupt and gradual
transition. It gives effective segmentation and so for
their used for video indexing and browsing. The
advantage of proposed method is that there is no use of
threshold for detecting. Therefore, algorithm is not
sensitive to content of video.
based on mutual information for better detection
accuracy and efficiency. In addition, the method employ
hierarchical analysis both spatially and temporarily on
the video data, leading to reduced computation and
improved gradual transition detection accuracy. Then
this algorithm is implemented on the public test video
data and results prove both the efficiency and the
accuracy of shot boundary detection.
Tuanfa Qin et al. have proposed a fast shot –boundary
detection based on K-step slipped window [10]. This
method considers the low-feature and edit-feature of the
video-shot. First, it selects the candidates of the shotboundary by K-step slipped window and adaptive
thresholds, and then, it does hard cut detection, flash
exclusion and gradual transition detection orderly. This
method compares its experiment results to some existing
method in terms of computational time and precision. It
has been found that this method has higher precision and
low computational time. It improves the efficiency and
precision of gradual transition detection
An accumulation algorithm for video shot boundary
detection has been proposed by T. Lu and P. N.
Suganathan [13]. The proposed algorithm takes the
difference between consecutive frames and accumulates
them. When accumulation difference exceeds a
threshold, transition is detected. The algorithm have
introduced C frame. The content of frame C represented
the changes from beginning of the shot. The frame C
takes the difference and similarity between first frame
and second frame i.e f1 and f2 and then compares with
consecutive frames until difference exceeds threshold.
As soon as it exceeds threshold, transition is detected.
This method can detects all shot boundary like cuts,
fades and dissolve. It resolved the problem of zooming
in the block matching methods.
Mohanta et al. have proposed model-based shot
boundary detection technique using frame transition
parameters [11]. Based on their proposed model,
formulate frame estimation scheme using the previous
and the next frames. The transition parameters along
with the error in estimation are determined from global
feature such as the color intensity histograms. Transition
parameters are also derived from local features like
scatter matrix of edge strength and motion matrix. The
transition parameters derived from global and local
features along with the error in probability density
function. Estimation describes the variability in visual
content of the neighboring frames. So the transition
parameters and corresponding errors constitute the
feature vector for shot boundary classification. Hence,
color intensity histogram, edge scatter nor motion
matrices are not directly used as features in frame
classification. Rather local variation (between frames) of
these statistics is used as features for shot detection and
classification. Thus feature set incorporates more
abstract information about shot transition, and at the
same time, the size of the feature vector is drastically
reduced. As a result, within shot variation and between
shot variations results are reflected at different degree.
For classification, they have employed a neural network
with back-propagation mechanism. It classifies the
frames into one of the three categories: no change,
gradual change, and abrupt change. Partial dissimilarity
measure and final shape of the decision boundary are
evolved to best suit the training data. Thus, the proposed
scheme is less dependent on the critical issues like
selection of various thresholds or sliding window size.
Finally, a simple but effective post processing is carried
out on the classified output to repair the false transition
and misclassification error, if any. The method is found
to be efficient and works better than the existing
algorithm for a variety of benchmark videos.
K. I. Koumousis et al. have proposed a new approach to
gradual transition detection [14]. In this method the
problem of shot boundary detection for gradual
transitions within video sequences is uses statistical tests
in conjunction with the Iterative Self Organizing Data
Analysis (ISODATA) classification algorithm for
consecutive video frames. The confusion matrix from
the classification results is formed in order to calculate
the Kappa coefficient and then it is used to identify the
transition.
A fast coarse-to-fine video shot segmentation algorithm
has been proposed by Liu and Jian-Xun Li [15]. The
camera motion, object motion and gradual shot
transition can be differentiated through this method.
Based on the improved information entropy theory, the
differences between the shots of a video sequence are
calculated. The adaptive thresholds are implemented to
select sequences of candidate shots from the video
sequence. Because the camera/object motion and the
gradual shot transition present similar characteristics,
they are all selected as the candidate shots. Then a fast
motion-edge detection algorithm is implemented to
distinguish the gradual shot transition. The proposed
algorithm is based on the statistical properties of the
characteristics, hence compared to the single
characteristic detection algorithm; the computational
complexity is reduced effectively. This algorithm is
reduces both the computational complexity and error
detections caused by the camera/object motion
effectively.
The comparative analysis of the various algorithms for
shot boundary detection are shown in Table I, in terms
of feature used, type of transition detected and their
merits.
Mutual information based video shot boundary detection
method has been proposed by Na Lv et al. [12]. The
proposed method improves upon the graph cut algorithm
__________________________________________________________________________
ISSN (Online): 2347-2820, Volume -1, Issue-2, 2013
4
International Journal of Electrical, Electronics and Computer Systems, (IJEECS)
_______________________________________________________________________
Using F1 measure, can rank the performance of the
different algorithms[4] . F1 combines recall and
precision with equal weight. F1 measure is a harmonic
average of recall and precision and is given below
V. EVALUATION CRITERION
For evaluation performance of shot boundary detection
algorithm, the two metrics recall and precision are used.
Recall is defined as
R=
C
C+M
=
C
D
F1( R, P ) =
(1)
C
(2)
C+FP
Where D is the total number of actual frames with
dissolves boundaries, C is the number of dissolve frames
correctly detected by the algorithm; M is the number of
number of dissolve frames missed by algorithm and FP
false positives detected by algorithm.
Algorithm
Using Eigen value,
decomposition and
Gaussian transition
detection
[1]
A fast shot detection
framework employing
pre-processing techniques.
[2]
Features used
Generalized
value
Eigen
Detection rate
DR=
Number of dissolves correctly detected (Hit)
Number of actual dissolve in the video ( Actual )
Types of transition
detection
Abrupt changes in hard cuts
and semi Gaussian behavior
in gradual transition.
3-DWT based motion
suppression
[6]
B-spline interpolation
[7]
Thresholding
and
bisection
based
comparison on large
no. of non boundary
frames.
Motion intensity and
motion
suppression
value
Low-scale and lowresolution image
DWT- based shot
boundary detection using
support vector machine
[8]
Color and the edge in
different
direction
from
wavelet
transition coefficients.
Using vector machines.
[9]
Color histogram and
χ2
Abrupt cuts and dissolve
K-step slipped window
[10]
Low-feature and editfeature of the videoshot
From global feature
such as the color
intensity histogram
Hard cut detection, flash
exclusion and gradual
transition detection
Classifies no change,
gradual change, and abrupt
change.
Model-based shot
boundary detection
technique using frame
transition parameters
[11]
(3)
R +P
However, the dissolve transition takes place over a
certain range of frames unlike gradual transition that
occurs at single frames. So, it is very difficult to
evaluate if the methods could actually identify different
types of dissolve transitions with camera and object
motion. Therefore, we need an additional evaluation
metric which can elaborate details about how many
dissolve types are correctly detected and is given by
Whereas precision is defined as
P=
2×R×P
Consequent hard cuts and
gradual transition.
Gradual
motion
transition
and
Classify the detected regions
into dissolve and fade types
by investigating the intra
frame standard deviation and
the number of consecutive
frames of solid Color
Cut transition, gradual
transition and normal
sequences.
(4)
Merits
Solve
the
problem
illumination and noise.
Significant speedup is
achieved.
Pricision and recall
satisfactory.
of
are
Solve the problem of
illumination and noise.
Undefined framework for
dealing with all gradual
transition.
Increases the effectiveness and
reduces the computational
cost.
Good result in recall with
arrange of 92.4-97.2%
No use of threshold for
detection
Not sensitive to content of
video.
Accuracy of shot detection.
Reduces the amount of
calculation.
Less dependent on selection
thresholds.
Effective post processing.
Reduces false transition and
misclassification.
__________________________________________________________________________
ISSN (Online): 2347-2820, Volume -1, Issue-2, 2013
5
International Journal of Electrical, Electronics and Computer Systems, (IJEECS)
_______________________________________________________________________
Mutual information based
[12]
An accumulation
algorithm
[13]
ISODATA classification
algorithm.
[14]
A fast coarse-to-fine video
shot segmentation
algorithm
[15]
Frame gray variance
based method and the
block color histogram
C frames
CUTs and GTs
Accuracy in shot boundary
detection.
Cuts, fades and dissolve
Kappa coefficient
Gradual effects
Effectively judges the small
motion in horizontal and
circular motion and optical
zooms.
Reduces the false detection
and improve the efficiency in
terms of precision and recall.
No need to have knowledge of
number of cluster
Difference between
DC images of all Iframe images
Camera motion, object
motion and gradual shot
transition
No use of threshold for
detection.
Not sensitive to content of
video
Table I. The comparative analysis of various shot boundary detection algorithm
The portion of the video considered for analysis consist
of 200 frames with camera motion (frames 16-217) and
75 frames with dissolve transition (frames 289-363).
VI. EXPERIMENTAL RESULTS:
We have selected the test video sequence which contains
significant camera motion with dissolve transition.
Fig.3 shows consecutive frames from the movie Xperia.
frame 16
frame 32
frame 54
frame 124
frame 187
frame 217
frame 289
frame 293
frame 301
frame 308
frame 315
frame 320
frame 333
frame 338
frame 352
frame 363
Fig.3. Consecutive frames from movie clip Xperia showing dissolve transitions
__________________________________________________________________________
ISSN (Online): 2347-2820, Volume -1, Issue-2, 2013
6
International Journal of Electrical, Electronics and Computer Systems, (IJEECS)
_______________________________________________________________________
Histogram difference between frames
12
10
8
6
4
2
0
0
50
100
150
200
250
Frame Index
300
350
400
Fig.4. Histogram difference between consecutive frames
We have applied color histogram difference between
frames as mentioned in [9] to the above test video
sequence. Fig.4 shows histogram difference of 370
consecutive frames. Here actual dissolve transition is
from frame 289-363, whereas camera motion is from
frame 16-217. It can be clearly observed from Fig, 4
that because of fast camera motion the histogram
difference between frames shows higher value than
dissolve transition. Hence, if we apply global or
adaptive threshold to the histogram difference between
frames, the algorithm will provide more false positives
than actual dissolve transition. If the shot boundary
algorithm falsely identifies camera motion as a shot
boundary then correspondingly key frames and video
retrieval will provide false results.
Algorithm has been presented in brief. The main focus
of the review is to detect gradual transitions in the
presence of camera and object motion. We feel that the
feature extraction techniques reviewed in this paper
will provide important clues to design efficient shot
boundary detection algorithm. After reviewing the
literature in details, we conclude that most of the
algorithms are unable to differentiate between gradual
transition and motion. In future, we wanted to develop
algorithm which will detect gradual transition without
showing any false positive for object and camera
motion in movie video.
REFERENCES:
[1]
Ali Amiri , Mahmood Fathy, “Video shot
boundary detection using generelized eigen
value decomposition and Gaussian transition
detection”, Computing and Informatics, vol. 30,
pp. 595- 619, 2011.
[2]
Y.-N, Li, Z.-M, Lu, X.-M, Niu, “Fast video shot
boundary detection framework employing preprocessing techniques IET, Image Process, vol.
3, iss. 3, pp. 121–134, 2009.
[3]
R. Bole, B. Yeo, M. Yeung, “Video query:
research directions”, IBM Journal of Research
and Development, vol . 42, iss. 2, pp. 233–252,
1998.
[4]
N. Dimitrova, H.-J. Zhang,
Sezan, T. Huang, A.Zakhor,
video content analysis and
Multimedia, vol. 9, iss. 3, pp.
Hence we required robust and effective algorithm
which can either suppress motion with respect to
dissolve transition or can differentiate between motion
and actual shot transition.
V. CONCLUSION
The demand for multimedia data services necessitates
the development of techniques to store, navigate and
retrieve visual data. The use of existing text indexing
techniques for image and video indexing is inefficient
and complex. Moreover, this approach is not generic,
and hence is not useful in a wide variety of
applications. Consequently, shot boundary detection
techniques should be employed to search for desired
images and video in a database. This paper reviews
shot boundary detection techniques proposed in the
recent literature. The main contribution of each
B. Shahraray, I.
“Applications of
retrieval”, IEEE
42–55, 2002.
__________________________________________________________________________
ISSN (Online): 2347-2820, Volume -1, Issue-2, 2013
7
International Journal of Electrical, Electronics and Computer Systems, (IJEECS)
_______________________________________________________________________
[5]
M. Lew, N. Sebe, P. Gardner, Video indexing
and
understanding, in: M. Lew (Ed.), “
Principles of visual information retrieval”,
Springer, Berlin, pp. 163–196; 2001.
[6]
Yang Xu, Xu De, Gaun Tengfei, Wu Aimin,
Lang Congyan, “3 DWT based motion
suppression for video shot boundary detection”,
Knowlwdge -based Intellingent Information and
Engineering System, Lecture notes in Computer
Science, vol. 3682, pp. 1204-1209, 2005
[7]
Jeho Nam and Ahmed H. Tewfik, “Detection of
gradual transitions in video sequences using Bspline interpolation” IEEE Transaction on
Multimedia, vol. 7, iss. 4, pp. 667-678, Aug
2005.
[8]
[9]
[10]
K-step slipped window”, 2nd IEEE international
Conference on Network Infrastructure and
Digita Content, pp. 190-195, 24-26 Sept 2010.
Jun Li, Youdong Ding, Yunyu Shi, Qingyue
Zeng, “DWT-based shot boundary detection
using support vector machine”, Information
Assurance and Security, vol. 1, pp. 435 – 438,
18-20 Aug 2009.
Vasileios Chasanis, Aristidis Likas, Nikolaos
Galatsanos, “Simultaneous detection of abrupt
cuts and dissolves in videos using support
vector machines”, Pattern Recognition Letters
30, pp. 55-65, 2009
[11]
Partha Pratim Mohanta, Sanjoy Kumar Saha,
and Bhabatosh Chanda, “A model-based shot
boundary detection techniqueusing frame
transition farameters, IEEE transaction on
Multimedia , vol. 14, iss. 1, pp. 223-233,
Feb.2012.
[12]
Na Lv, Zhiquan Feng and Jingliang Peng,
“mutual information
based video shot
boundary detection” Image Analysis and Signal
Processing, pp. 1-5, 9-11 Nov. 2012
[13]
T. LU tong, P.N. Suganthan, “An accumulation
algorithm for video shot boundary detection”,
Multimedia Tools and Applications,vol. 22, iss.
1, pp. 89–106, Jan 2004.
[14]
K. Koumousis , V. Fotopoulos, A. N. Skodras ,
“A new approach to gradual video transition
detection”, Informatics(PCI), pp. 245-249, 5-7
Oct 2012
[15]
Liu Liu, Jian-Xun Li, “A Novel Shot
Segmentation Algorithm Based on Motion Edge
Feature”, 2010 Synopsis on Photonics and
Optoelecctronic, pp. 1-5, 19-20 June 2010.
Tuanfa Qin, Jiayu Gu, Huiting Chen, Zhenhua
Tang, “A fast shot boundary detection based on

__________________________________________________________________________
ISSN (Online): 2347-2820, Volume -1, Issue-2, 2013
8