融合背景差分法和帧间差分法的运动目标检测

Moving Target Detection Using Background Difference Method and Inter-Frame Difference Method

针对传统背景差分法容易受光照变化的影响和传统帧间差分法不能完整地检测出前景目标等问题,提出一种改进的将两种算法进行融合的算法。首先,对背景差分法,采用带光照补偿模型的均值背景建模,降低光照突变对目标检测有效性的影响;然后,对帧间差分法,采用基于最大模糊熵阈值的三帧差分对视频图像进行处理,降低检测目标轮廓不完整的概率;最后将两种改进的算法得到的前景目标进行融合,得到综合的前景目标。实验证明,该算法能有效弱化光照突变的影响,增强检测的准确性和实时性。

Aiming at the problems that the traditional background difference method is susceptible to changes in illumination and that the traditional inter-frame difference method cannot completely detect the foreground target,proposes an improved algorithm that combines the two algo⁃rithms.First,for the background difference method,the mean background modeling with the illumination compensation model is used to re⁃duce the impact of light mutation on the effectiveness of target detection.Then,for the inter-frame difference method,three frames of differ⁃ential pair video based on the maximum fuzzy entropy threshold The image is processed to reduce the probability of incomplete target con⁃tour detection.Finally,the foreground targets obtained by the two improved algorithms are fused to obtain a comprehensive foreground tar⁃get.Experiments show that this algorithm effectively improves the detection performance of moving targets and has better adaptability to sudden changes in illumination.