Object tracking and navigation based on computer vision / Somaia Mohamed Mahmoud Mohamed Mohamed ; Supervised M. Fekri , M. Hesham

Thesis (Ph.D.) - Cairo University - Faculty of Engineering - Department of Electronics and Communications

This work proposes novel modify on algorithm for two different applications in computer vision namely localization based on vision and object tracking in video. The proposed work in this thesis is intended to improve the conventional system for object tracking through using wavelet transform and morphological operations. The new algorithm succeeds in removing the noise and smooth the image, as well as scale invariant feature transform (SIFT) to increase the number of features which, in turn, improves the matching stage. The proposed procedure has been applied on many consequent frames, under different circumstances. The implementation results show high accuracy and performance, compared with the other conventional procedures such as Wavelet-based object tracking algorithms, and SIFT-based object tracking algorithms. On the other hand, the proposed navigation algorithm determines the location of the moving objects by using a computer vision system. A single camera system is used to select suitable features and a stereo camera system to obtain the locations of the object. This significantly improves the overall accuracy, compared with the conventional systems that mainly depend on maps to determine the location of the object by comparing the features of the image with the features of the map. The implementation results of the proposed object localization show high performance that achieved 10 meters{u2019} accuracy on the trajectory with a length of 165 meters

Issued also as CD