Abstract:
Discovery of a moving object in a video is simple if the image sequences are stable but will be hard if the captured images are spatially and temporally deformed. It is even more difficult if the target is just few pixels in size. Moreover, motion adjustment is a key preprocessing step in dynamic image analysis, which deals with the removal of unwanted motion in a video sequence and if the objects of interest are additionally moving in the scene, their motion will be mixed up due to different atmospheric conditions in the captured images, rendering the problem of detecting the moving objects extremely difficult. Asynchronous motion adjustment, target discovery and tracking is a key problem and become more challenging when it is under violent atmospheric conditions. Our goal is to detect true target in a corrupted video with tracking record having static background and analyzing target behavior. Due to rapid development in technology, wide numbers of techniques have been proposed to deal with motion adjustment, target discovery and tracking separately but asynchronous approach is rare. So there is a need for efficient, cost effective and reliable asynchronous approach for motion adjustment, target discovery and tracking.
In this thesis, a robust motion adjustment, target discovery and tracking approach is proposed for different surveillance purposes. The proposed approach utilizes the concept of mean and standard deviation for exact detection of true objects. After getting true objects and their tracking record the behavior of object is analyzed by movement estimation of its centroid. Further an adjusted background estimated by relative smoothness and running average techniques.
Experimental results show that outlier free true targets are detected in the videos corrupted by violent atmospheric conditions. Target tracking record and action state with static background reveals the efficiency, reliability and robustness of this approach. The proposed approach can be used for different surveillance purposes.
