静态手势图像轮廓特征提取算法研究

Study on Extraction Algorithm for Static Hand Gesture Contour Feature

在手势识别过程中,手势特征的提取非常重要,如果提取的手势特征不具备较好的可分辨性和表征的不变性,就很难达到手势识别的要求。由于人手经常会弯曲,手指也常会被手部其他位置遮挡,再加上所在环境光照的影响手势图像会出现高亮区域和阴暗区域,使得在设定初始轮廓曲线与手势轮廓较远时手势分割出现手势区域有漏检的情况,而且在手势轮廓凹形区域不易识别等问题,导致同样的手势得到不同的手势轮廓描述,影响手势识别的可靠性,为此本文提出了一种主动轮廓与肤色统计融合的静态手势轮廓特征提取算法解决这一问题。

In the process of gesture recognition,the gesture feature is very important. If the extracted gesture feature does not have good distinguishability and the feature invariance,it is difficult to meet the requirement of gesture recognition. Because the hand is often bending,finger is also often shaded by other part of hand and with the influence of environmental light,the gesture image will have a highlight areas and dark areas. So in setting the initial contour curve or when the gesture contour is far away,the situation occurs that the gesture area is missing when making gesture segmentation. And the problem that the region of gesture concave contour is not easy to be identified can lead to that the same gesture gets different gestures outline description and affect the reliability of gesture recognition.This paper proposes a static gestures contour feature extraction algorithm which integrating the active contour and the color statistic to solve this problem.