复杂背景下的人体轮廓提取算法

Human Body Contour Extraction Algorithm under Complex Background

针对现有人体轮廓提取方法在复杂背景情况下鲁棒性差、精确度不高的问题,提出了一种改进的自适应人体区域分割方法。首先提取肤色样本,并建立自适应高斯模型提取肤色区域,然后在HSV色彩空间分割提取服装区域,最后线性融合以上两检测区域实现人体区域的检测。实验结果表明,该方法可以得到比较准确且完整的人体区域,有效降低了复杂背景环境、光照等条件的限制。

Aiming at the problem that the existing human body contour extraction method is poor in robustness and low precision under complex background,an improved adaptive human body region segmentation method is proposed.First,the skin color samples are extracted,and an adaptive Gaussian model is established to extract the skin color region.Then,the clothing areas are extracted in the HSV color space.Finally,the above two detection areas are combined to realize the detection of the human body.The experimental results show that the method can obtain a more accurate and complete human body region,which effectively reduces the limitations of complex background environment and illumination conditions.