MLCMRF

In this work, we propose a multi-layer Compound Markov Random Field (MLCMRF) Model to spatially segment the different image frames of a given video sequence. The proposed MLCMRF uses five Markov models in a single framework, one in spatial direction using color feature, four in temporal directions (using two color features and two edges/line fields). Hence, the proposed MLCMRF is a combination of spatial distribution of color, temporal color coherence and edge maps in the temporal frames. The use of such an edge preserving model helps in enhancing the object boundary in spatial segmentation. The segmented image frames are combined with the gray level difference based change detection mask (CDM) to detect the moving objects from other image frames. and hence can detect moving objects with less effects of silhouette. The MAP estimate of the considered MLCMRF is obtained by a combination of simulated annealing (SA) and iterative conditional mode (ICM), which converges fast. A gray level difference based change detection mask (CDM) is constructed and is subsequently updated with the previous frame video object plane (VOP) and the spatial segmentation of the consecutive frames, to detect the moving

objects from the target image frames.

Publication

1. B. N. Subudhi & P. K. Nanda, “Compound Markov random field Model based video segmentation,” Proceedings of SPIT-IEEE colloquium and International conference 2007- 2008, vol.1, pp. 97-102, 2008. (pdf)

2. B. N. Subudhi & P. K. Nanda, “Moving Object Detection Using Compound Markov random Field Model,”Proceedings of IEEE Conference Computational Intelligence, Control and ComputerVision in Robotics and Automation (CICCRA-2008), vol.1, pp.198-204, 2008 . (pdf)

In this scheme, segmentation is considered as a pixel labeling problem and is solved using the maximum a' posteriori probability (MAP) estimation technique. The MRF-MAP framework is computation intensive due to random initialization. To reduce this burden, we propose a change information based heuristic initialization technique. The scheme requires an initially segmented frame. For initial frame segmentation, compound MRF model is used to model attributes and MAP estimate is obtained by a hybrid algorithm (combination of both Simulated Annealing (SA) and Iterative Conditional Mode (ICM)) that converges fast. For temporal segmentation, instead of using a gray level difference based change Detection Mask (CDM), we propose an use of CDM based on label information based difference of two frames. The proposed scheme resulted in less effect of silhouette.

Publication:

1. B. N. Subudhi & P. K. Nanda, “An evolutionary based slow and fast moving video object detection scheme using Compound Markov Random Field Model,” Proceedings of IEEE - 6th Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP-08), vol.1, pp. 398-405, Dec 2008. (pdf)

2. B. N. Subudhi, P. K. Nanda & A. Ghosh, “A change information based fast algorithm for moving Object Detection and Tracking,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 21, no. 7, pp. 993-1004, 2011. (pdf)

In this work, a local histogram thresholding scheme is proposed to segment the difference image by dividing it into a number of small non-overlapping regions/windows and thresholding each window separately. The window/block size is determined by measuring the entropy content of it. The segmented regions from each window are combined to find the (entire) segmented image. This thresholded difference image is called the change detection mask (CDM) and represent the changed regions corresponding to the moving objects in the given image frame. The difference image is generated by considering the label information of the pixels from the spatially segmented output of two image frames. The generated CDM is combined with the segmentation of MLCMRF output for object detection.

Publication:

1. B. N. Subudhi, P. K. Nanda & A. Ghosh, “Moving Object Detection Using MRF Model and Entropy based Adaptive Thresholding,” Proceedings of IEEE 2nd International Conference on Human Computer Interaction (published by Springer), 1:155–161, 2010. (pdf)

2. B. N. Subudhi, P. K. Nanda and A. Ghosh, "Entropy based region selection for local thresholding to detect moving objects", Pattern Recognition Letters, vol. 32, no. 15, pp. 2097-2108, 2011. (pdf)

In this model we have proposed a multi-layer markov framework, it takes care of spatial distribution of current frame, temporal frames and the Change Detection Masks (CDM) of the temporal frames.

Publication:

1. B. N. Subudhi & P. K. Nanda, “Detection of Slow Moving Video Object using Compound Markov Random Field Model,” Proceedings of IEEE- TENCON 2008, Hyderabad, 2008. (pdf)