A background subtraction algorithm based on biological vision characteristics

For the problem of how to build a robust background model and update the background model, a background subtraction algorithm based on biological vision characteristics combined with the ViBe algorithm was proposed. Firstly, utilizing the extrinsic nearsightedness characteristic of the frogs visual system, the meaning of a pretreatment method called “region fuzzy” and its implementation were described. Then, considering the characteristics of color cognition by humans, a measurement criterion for color difference based on Webers law in the LUV color space was given. Finally, specific implementations of the algorithm were introduced from three main aspects: Background modeling, foreground detection and background model updating. Experimental results show that this algorithm can improve the accuracy of moving object detection.