Improved Visual Background Extractor Model for Moving Objects Detecting Algorithm

As visual background extraction model is not adaptable enough for dynamic scenes and provides with low accuracy in moving target detection,an improved model of the visual background algorithm is proposed in this paper.In model initialization,neighborhood background model is established by adding the sequentially extracted pixel information to bring down the impact of complex scenes.In order to reduce the interference of complicated scenario on the test results and improve the accuracy of motion target detection,the segmentation threshold is adaptively obtained by combined with the spatial neighborhood information.To enhance the robustness and adaptability in complex scenes,the update cycle and update method of the model are dynamically adjusted according to the scene complexity in the process of background update.The influence of ghost and background noise can be effectively eliminated by updating.Experimental results show that the improved algorithm is applicable to moving object detection in dynamic scenarios and provides with higher detection accuracy.