Robust and Efficient Transform Domain Video Sequence Analysis: An Approach from the Generalized Color Ratio Model

Adjeroh, Donald A.; Lee, M.C.

doi:10.1006/jvci.1997.0349

Cited by 16 publications

(22 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is often the case in the analysis of remote sensing imagery or analysis of video/image sequences [1], [2]. The paper did not consider the histogram generation process, nor did it discuss the related issue of index structures.…”

Section: Discussionmentioning

confidence: 99%

“…We refer to this internal statement of the required performance level as the allowed probability of random match. Typically, the user may be given a range of values, (e.g., 2 In this paper, we use image objects and visual data objects interchangeable. , where zero is the minimum level and is the maximum performance level, we use the following mapping to relate the minimum acceptable level of performance , (indicated by the user), and the internal probability of random match, :…”

Section: A Performance Guaranteesmentioning

confidence: 99%

“…Then, there are possible ways of placing the events into cells. Assuming each of the ways are equally probable, we can use the binomial distribution to obtain the probability that a given cell contains exactly events: (2) For the normalized histogram, we know and , but may not know at the time of matching. (Typically, depends on the specific feature used.)…”

Section: B Formal Thresholdsmentioning

confidence: 99%

See 2 more Smart Citations

An occupancy model for image retrieval and similarity evaluation

Adjeroh

Lee²

2000

IEEE Trans. on Image Process.

Self Cite

View full text Add to dashboard Cite

The analysis of visual information often involves the manipulation of enormous volumes of data. If some tolerance is allowed in the results, orders of magnitude improvement in efficiency can be achieved in such analysis by appropriate selective processing, without necessarily considering all the data features. To guarantee that the error introduced does not exceed the allowed limit, a certain minimum proportion of the data must be involved in the analysis. This proportion cannot be determined arbitrarily. It should be chosen based on some formal methods, with a consideration of the error inherent in the data. This paper presents some techniques for improving the retrieval efficiency in image-based information systems, with performance guarantees on the reliability of results. Using the statistical theory of occupancy, it develops a model for the formal selection of the minimal subset of image features to be involved in histogram-based similarity evaluation. This guarantees that decisions based on the minimum proportion are always the same as (or close to) the one that would have been reached by considering all the features. Results on real and simulated data show the performance of the model on speedup, robustness, scalability, and performance guarantees.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: A Performance Guaranteesmentioning

confidence: 99%

Section: B Formal Thresholdsmentioning

confidence: 99%

See 1 more Smart Citation

An occupancy model for image retrieval and similarity evaluation

Adjeroh

Lee²

2000

IEEE Trans. on Image Process.

Self Cite

View full text Add to dashboard Cite

show abstract

“…[3] had used model based comparison which considered the video production system as a template. [4] used an Edge features for detection shot boundary where it looks for entering and exiting edge pixels.…”

Section: IImentioning

confidence: 99%

Shot Boundary Detection Using HSV Colormap for Content Based Video Retrieval

Shah¹

2017

IJRASET

View full text Add to dashboard Cite

I.INTRODUCTION Rapid developments of digital video capturing and editing devices and technology have been producing an ample of digital video data every day. As a consequence, digital video data information keep on increasing everyday and so need of an efficient application arrives; such as video-on-demand or content based video retrieval. This technology of how to catalogue and retrieve videos on demand for future re-use becomes a need of society. In automatic management of video database, it is practically inappropriate to use keywords to describe each video sequence because this annotation process asks for enormous human power; in addition keywords used cannot describe the subject properly. Hence, to fulfil this demand of retrieving similar videos from the database, we need a system Content based video retrieval which retrieves the same videos based on video content analysis. Primary tasks of CBVR system are shot boundary detection, key-frame extraction. Shot boundary detection becomes the groundwork for video retrieval and management. Shot boundary detection is allowing the computer to discover the editing positions and produce the original shot sequences. The detected shots will become the basic query units in video retrieval systems. Video is a space and time varying structure captured by either single camera or multiple cameras. Video structure can be divided as video scenes, video shots, video frames and video key frames. All play a major role for video indexing and video retrieval applications. A video shot is a succession of frames captured by a single camera in a continuous run. Video shot is a group of sequence of similar action frames which carries almost same information and visual features such as colors, motions and textures. The visual information of each shot of the video can be described by one or multiple frames, called key-frames. The number of keyframes cannot be predetermined because of content variation. In case of a static shot there is little content variation, so a single keyframe can describe the whole shot effectively, whereas in case of a high camera editing and object motion shot, we may need more key-frames for a better representation of a video content (Vasileios T. Chasanis Jan. 20093). Video shot boundary detection is a primary step in video indexing and retrieval, and in general video database management. Shot boundary detection is used to segment a given video into its continuous shots in a way it was produced, and to classify and detect the different shot transitions (Don Adjeroh May. 20094). Different algorithms have been proposed, based on color histograms (Ballard 1991, Tanaka 19925), pixel color differences (H. Zhang 1993), color ratio histograms (Lee 1997), edges (R. Zabih 1999), and motion. In this proposed work, we studied the problem of video shot classification and separation using statistical features of HSV color map without calculating statistical threshold. Manual change of threshold is not a practical solution for shot boundary detection. Proposed approach does...

show abstract

“…Shot detection techniques can also be classified into decompressed video detection and direct compressed video detection. The second approach typically takes advantage of coding information in compressed video to determine the location of shot transition and is expected to be more efficient as it does not need to decompress the video [19,20].…”

Section: Related Workmentioning

confidence: 99%