复杂背景下的人耳检测方法

Ear Detection Method in Complex Background

本文在借鉴人脸检测算法的基础上,对现有的人耳检测方法进行研究,针对人耳区域小、共性特征少、复杂背景下难以检测等特点,提出了一种分阶段优化的静态彩色复杂背景下的人耳检测方法。首先采用肤色分割将检测范围缩小至肤色区域;接着利用侧脸的先验知识再次筛选;然后对图像进行边缘提取,在边缘图像中利用人耳含有丰富边缘信息的特点进行区域搜索以确定人耳。在该方法中,彩色图像的肤色信息和灰度图像的多尺度边缘以及人耳自身特征被结合起来,对解决人耳共性特征在复杂背景下难以被提取的问题具有较好的效果。实验结果表明,该算法在复杂背景下是有效的。

Based on the research of human-face detection, existing ear-detection methods are investigated in this paper, and a new optimized detecting method is presented to solve difficulties in ear detection such as small region, few common features, complex background and so on. Firstly, images are segmented into different regions with skin-color information and potential candidate regions are selected. Secondly, prior knowledge of side-face is applied to eliminate impossible regions. Thirdly, image edge is detected in different scales and ears are detected according to the rich edge information of human ear. In this method, the skin information of color image and multi-scale edge of grey image are combined with human ear features, which are employed to solve the problem that common features of ear image are difficult to extract in complex background. Experimental results show that this method is effective in complex background.