摘要
针对复杂光照与背景的图像中难以准确检测出人脸的情况,提出一种复杂光照与背景下的Double肤色模型人脸检测算法。采用一种改进的光照补偿方法对图像进行光照预处理,融合改进的高斯肤色模型和非线性肤色模型,通过对融合算法进行收敛性验证与实例检验,证明了其可行性。实验结果表明,该算法能够对复杂背景及光照下多人脸、多姿态、遮挡、动态视频图像等情况的人脸检测有较好的检测效果。
Focused on the accurately detect human face in the image of complex illumination and background, this paper proposes a face detection algorithm of a double color model under the condition of complex illumination,uses an improved illumination compensation algorithm for illumination preprocessing and integrates the improved Gaussian color mode and the Nonlinear color model. By verifying the astringency,we use an example to test its feasibility. The experimental resuhs show that the proposed algorithm can achieve good effect on face detection of muhiplayer face, multi-posture, occlusion, dynamic video images under the condition of complex background and illumination.
出处
《江南大学学报（自然科学版）》
CAS
2014年第5期517-522,共6页
Joural of Jiangnan University (Natural Science Edition)
基金
国家自然科学基金项目(61170277)
上海市教委科研创新重点项目(12ZZ137)
关键词
光照补偿
Double肤色模型
复杂光照
多姿态人脸检测
illumination compensation, Double color model,complex illumination, multi-posture face detection
