【摘要】 GMM算法在目标检测中采用固定的模型个数描述像素点的状态，固定的学习率更新背景。针对GMM算法的以上不足，提出一种自适应选取模型个数和学习率自适应环境变化的GMM算法。通过模糊理论将视频分为三个模糊子集，计算每一部分的模糊熵，根据视频模糊熵的最大值确定需要的模型个数。引入两帧视频的相关性度量视频帧之间的相关性，比较背景变化因子与背景变化系数确定不同的学习率。实验验证，改进算法能够消除噪声影响，有效地节约混合高斯模型的个数，能够自适应环境变化，提高检测准确率，降低检测耗时。更多还原

【Abstract】 The GMM algorithm uses a fixed number of models in the target detection to describe the state of the pixels and a fixed learning rate to update the background. In view of the above shortcomings of the GMM algorithm, a GMM algorithm that adaptively selects the number of models and the learning rate to adapt to environmental changes is proposed. The video was divided into three fuzzy subsets by fuzzy theory, then the fuzzy entropy of each part was calculated, and the required number of models was determined according to the maximum value of the fuzzy entropy of the video. The correlation between two frames of video was introduced to measure the correlation between video frames, and the background change factor and background change coefficient were compared to determine different learning rates. Experiments verify that the improved algorithm can eliminate the influence of noise, effectively save the number of mixed Gaussian models, and can adapt to environmental changes. The detection accuracy is improved, and the detection time is reduced.更多还原