摘要
提出了一种应用于彩色图像的改进的Snake眼睛边缘检测算法。该算法首先将输入图像从RGB空间转换到YCbCr和YCb′Cr′空间,依据肤色点在Cb′Cr′空间中的聚类性提取人脸候选区域;然后通过在人脸候选区域上生成眼睛图(EyeMap)提取眼睛候选区域并计算候选区域Cr2图像的距离势能力场。采用两段首尾相连的抛物线进行初始化并从外部对Cr2图像的距离势能力场进行活动轮廓模型(Snake)匹配,在匹配过程中不断进行修正使检测结果更接近真实边缘。实验结果证明,该算法能精确地检测出眼睛的边缘并节省匹配时间。
In order to detect eye edge accurately, an improved eye edge detection algorithm for color image is proposed in this paper, Based on color transform from RGB space to YCbCr space and YCb′Cr′ space, the algorithm detects skin regions over the entire image and generates face candidates in term of the spatial arrangement of these skin patches in Cb′Cr′space, Then, an EyeMap is built to generate eye candidates. The distance force field of Cr^2 images of eye candidate is used for active contours models （Snake） deformation, and a correction procedure is constructed during deformation, Two end-to-end parabolas outside the candidate are used for initialization and Snake matching. Experimental results show that the proposed algorithm can detect eye edge accurately and save matching time.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2007年第10期78-82,133,共6页
Opto-Electronic Engineering
