Abstract: Moving objects can be converted into background ones by Gaussian Mixture Model(GMM) when they might be staying still in the scene for an uncertain time. Therefore, a new method based on improved GMM is proposed. To obtain complete objects, a background learning parameter is introduced to update the model according to the detection result in previous frame. Moreover, the information of 8-adjacent connection area is utilized to suppress noises and improve its stability in the complex environment. Several experiments are implemented and the results demonstrate its effectiveness in detecting the moving objects which stay briefly in the complicated condition.

Key words: Gaussian Mixture Model(GMM), moving object detection, background model, noise suppression, connection area

摘要： 在运动目标停滞的情况下，传统的混合高斯模型会将运动目标误判为背景，导致目标漏检。为此，提出一种基于改进混合高斯模型的目标检测算法。引入背景学习参数，结合前一帧的检测结果自适应地更新背景，从而提取完整的运动目标。利用像素的八连通区域信息抑制噪声，提高算法在复杂环境中的稳定性。实验结果表明，与传统检测方法相比，该算法能够在复杂环境中准确地检测出短暂停滞的运动目标。

关键词: 混合高斯模型, 运动目标检测, 背景模型, 噪声抑制, 连通区域

CLC Number: 

• TP391