Moving Object Detection Algorithm Based On Background Modeling In Dynamic Scene

Object detection occupies an important position in the field of computer vision.The paper describes overseas and domestic research actualities of video surveillance system.We analyze the role and status of background modeling algorithms in moving object detection and compare the performance of classical background modeling algorithms.In addition, we put forward a novel moving object detection algorithm based on enhanced dual model algorithm and a novel background modeling algorithm based on fusion of textures and color features.In the paper, we study the theories of background modeling and elaborate the advantages and disadvantages of four classical background modeling algorithms.Moreover, the detection results and applicable scenes are verified as well.The paper presents an improved dual model moving target detection method as the result of the misjudge rate of the existing dual model algorithms, incompleted object detection and appearance of ghost in dynamic scenes. Firstly, the way of background judgement is improved to distinguish foreground comprehensively. The distance between the new pixel value and the sample value of background model is compared with the threshold. Secondly, according to the result of the foreground and background of the self model and the neighborhood model, as the condition of the increase and decrease of the threshold value, the adaptive threshold update method is improved. Finally,by comparing the corresponding change of position pixel value to determine the ghost pixel,the paper combines with the inter frame difference technique in order to achieve the purpose of eliminating ghost rapidly.We propose the background modeling algorithm based on texture features and color features in the paper .Firstly, extracting color features whose RGB three channel value is normalized and the texture energy features which is based on the region.Secondly,the background model is constructed by two kinds of features;Finally,according to the dynamic feedback mechanism of complex background, the threshold of adaptive clustering and adaptive update threshold are obtained for random clustering.The experimental results show that the algorithm has obvious advantages in the dynamic scene.