Research Of Object Tracking Algorithm Based On Color Space Nonuniform Quantization And Multi-information Fusion

Motive object tracking is the core research content in the field of computer vision,aimed at seeking for the interest object in each frame of video or sequential images,according to the specified initial information of target.This technology is extensively applied in human-computer interaction,human behavior analysis,object detection,video retrieval,video surveillance,virtual-reality,tele-medicine treatment,etc.In recent years,the object tracking algorithm has been developed and updated under the unremitting efforts of scientific research personnel.According to the principle of object tracking algorithms,they are usually divided into four classifications,tracking based on region,tracking based on active contour,tracking based on model,tracking based on characteristics.Currently,the existing object tracking algorithm is still hard to deal with background interference,target occlusion,target appearance drastic changes,multi-information fusion,etc.There is still much room for improvement.In allusion to the limitation of object tracking algorithm at this stage,the main work in this paper are as follows:Firstly,we analyzed the research background and significance of object tracking technology,learned the principle of the existing four classifications of object tracking algorithm and characteristics of target which are commonly used in tracking.Secondly,in allusion to the problem of background interference,an optimized CAMshift algorithm is proposed.To improve the target-background distinction degree and suppress the interference of background pixels,the component of brightness and saturation are introduced in histogram of classic CAMshift algorithm,and the color space is nonuniform quantized.Thirdly,in allusion to the problem of multi-information fusion,a asynchronous fusion algorithm within the forward prediction framework based on out-of-sequence measurements is proposed.Within the forward prediction,a optimal OOSMs fusing process is concluded,a information filter based on Equivalent Measurement is proposed,it could solve the problem of arbitrary-step-lag OOSMs.Finally,to verify the validity of algorithm proposed in this paper,we compared the simulation results of proposed algorithms and existing algorithm which get from Matlab simulation software.