Research On Image Stitching And Tracking Technology In Video Surveillance

The image stitching technology and target tracking technology are usually used to obtain the large surveillance scope and continuous trajectory of the target in the video surveillance.However,target tracking technology and image stitching technology based on feature point matching method and image fusion method,have stringent requirements in the illumination variation,noise,occlusion,and motion blur.Therefore,the thesis research on the above three methods and the main contents are as follows:The traditional ORB method has the high false matching rate under the influence of image noise,therefore,the feature point matching method based on distance fusion is proposed.SURF algorithm is used to extract feature points,and BRIEF method is used to describe feature points.The polarity and the amplitude information of random point pairs in the neighborhood of the feature points are simultaneously used to match the feature points.Two Hamming distances are fused by the adaptive weighted to measure the similarity distance of feature points.Compared with the traditional matching method,the improved method could get the higher correct matching rate,and the matching results could be used to improve the scene stitching performance.The traditional gradual fusion algorithm cannot eliminate the stitching gap caused by the uneven local brightness of the image,so an improved gradual fusion algorithm is proposed.The values of pixels in the overlapping area based on two images are fused by the nonlinear weights to eliminate stitching gaps.Compared with the traditional gradual fusion algorithm,the improved method not only could improve the effect of image fusion significantly.The kernel correlation filter tracking algorithm cannot precisely locate the target position due to the illumination variation,occlusion and motion blur,an improved tracking strategy is proposed.A new Histogram of Hue Gradient(HHG)feature is designed,and the new HOG-HHG feature is obtained by connecting the HOG and the HHG in series.Two features,CN and HOG-HHG,are extracted respectively,and two kernel correlation filter classifiers are constructed base on the two features above to establish the corresponding response maps of the tracking scenes,respectively.The response maps are fused adaptively to improve the tracking robustness to the complex situations in the tracking process.The updating strategy of the target model is designed based on PSR and its difference to improve the stability of the target model.Simulation results show that the proposed method has better tracking adaptability to the illumination variation,occlusion and motion blur.Both the precision and the success rate could be improved.The surveillance system based on the speech interaction platform was constructed.According to the above research contents,the compound tracking strategy is proposed for continuous target tracking in multiple surveillance scenes.Meanwhile the strategy is applied to the system to realize the functions of automatically selecting target,broadcasting lost content and continuous tracking.