结合背景更新和亮度范围的改进Codebook模型算法

摘要

Traditional Codebook background modelling algorithm is established in RGB colour space,it has the problems of complex calcu-lation,low segmentation quality and poor anti-jamming capability.In this paper,we first built the Codebook background model by utilising the characteristics of luminance and chrominance separation of YUV colour space.Secondly,we used the mean of Y component of codeword in Co-debook to redefine brightness range aimed at the problem of unreasonable definition in brightness range of background model.Finally,in the process of moving object detection we employed the background update method combining two-layer Codebook and short-sliding window to im-prove anti-jamming capability.Experiments showed that the brightness range in improved algorithm can update gradually with the background changes.Compared with traditional Codebook background modelling algorithm,this one can effectively update background and meanwhile im-prove detection accuracy as well.%传统的 Codebook 背景模型算法是在 RGB 颜色空间下建立的，该算法计算复杂、分割质量不高且抗扰动能力较差。首先利用 YUV 颜色空间下亮度与色度分离的特性，建立 Codebook 背景模型，其次针对背景模型的亮度范围定义不合理的问题，采用码本中码字的 Y 分量均值重新定义亮度范围，最后在运动目标检测过程中使用双层码本与短时滑动窗口相结合的背景更新方法提高抗干扰能力。实验表明，改进算法的亮度范围能够随背景变化不断更新。与传统的 Codebook 背景模型算法相比，该算法能够有效更新背景，同时提高检测的准确性。