基于Choquet模糊积分的运动目标检测算法

A Moving Object Detection Algorithm Base on Choquet Integrate

本文提出了一种基于Choquet模糊积分的运动目标检测算法(CIMOD,Choquet Integrate-Based Moving Ob-ject Detection).将模糊测度和模糊积分理论应用于运动目标与背景分类中,提出了自适应阈值的Choquet积分算法,实现了图像的颜色特征和纹理特征相融合;选择YCbCr颜色空间代替传统RGB空间,将图像亮度与色度分离,降低了光照变化对运动检测的影响;利用局部二元模式(LBP,Local Binary Pattern)纹理特征对亮度级的单调变化具有不变性的特点,将其融合到检测算法中,有效抑制了阴影的干扰.仿真实验结果表明,即使在光照变化、阴影干扰等复杂背景情况下,该算法也能够准确的检测出运动区域.

A moving object detection algorithm is presented, which base on Choquet fuzzy integral. This algorithm classifies the moving object and background using the fuzzy measure and fuzzy integrate theory, proposes a Choquet integrate algorithm with adaptive threshold to integrate the color features and texture feature; Chooses YCbCr color space instead of RGB space, which space can separate the intensity and the hue of image, to reduce the influence of illumination changing; introduces the Local Binary Pattern （LBP） texture feature to the detection algorithm, which is invariant to monotonic changes in pixel value scale, efficiently inhibits the shadows＇ disturbance. Experiment results indicate that background can be subtracted correctly by using this new algorithm, even ff the complex background with illumination changing and shadows.