一种基于混合高斯模型的运动目标阴影检测策略

摘要

The object division and shadow detection technology of video is one of key research directions in computer vision .A shadow elimination strategy was put forward based on GMM .First, suspected shadow was decided by colour angle under the HSV color model , then the moving targets together with suspected shadow were modeled with GMM so as to eliminate the true shadow .The experi-ments suggested that the strategy was capable of detecting and eliminating the shadow accompanied with moving targets which in envi -ronments with different light .The experiments results demonstrated the good robustness of the strategy that would ensure continuity and accuracy during targets detection at the same time .%视频的目标分割与阴影检测技术是计算机视觉领域中最主要的研究方向之一。基于混合高斯模型，提出一种双重阴影检测策略。先通过HSV模型下的颜色夹角确定疑似阴影，再对运动目标和疑似阴影进行混合高斯建模从而消除实际阴影。实验结果表明，该策略在不影响目标识别的情况下可以较好地检测并消除在不同光照环境下的随伴运动目标阴影，具有较高的鲁棒性，从而能保证目标检测的连续性和准确性。