面向静态手势识别的边缘序列递归模型算法

发布时间：2018-05-19 06:52

本文选题：手势识别 + 边缘序列　；参考：《计算机辅助设计与图形学学报》2017年04期

【摘要】：针对传统手势识别准确率不高、鲁棒性不强的问题,通过研究静态手势轮廓特征,从手势边缘序列角度出发提出一种基于手势边缘轮廓递归图的CK-1距离的手势识别算法CSRP.首先通过阈值分割获取手势区域图像;然后定位起始点坐标,建立随着空间位置变化的手势边缘序列;为了克服边缘序列数据的不等长问题,构造基于时空域的手势轮廓序列递归图;最后利用MPEG-1压缩算法计算手势递归图像之间的CK-1距离,完成手势识别.实验结果表明,该算法在手势发生旋转、平移、缩放时具有较高的鲁棒性,并且计算量小、效率高,手势识别的准确率高达97%.
[Abstract]:Aiming at the problem that the accuracy and robustness of traditional gesture recognition are not high, a gesture recognition algorithm based on CK-1 distance based on gesture edge recursive graph is proposed by studying the static gesture contour features and starting from the angle of gesture edge sequence. First, the gesture region image is obtained by threshold segmentation; then, the starting point coordinate is located, and the gesture edge sequence with spatial position is established. In order to overcome the unequal length of edge sequence data, Finally, the MPEG-1 compression algorithm is used to calculate the CK-1 distance between the gesture recursion images to complete the gesture recognition. The experimental results show that the algorithm is robust when the gesture is rotated, translated and scaled, and the computation is small, the efficiency is high, and the accuracy of gesture recognition is as high as 97%.
【作者单位】：西北大学信息科学与技术学院;
【基金】：国家自然科学基金(61305032) 陕西省教育厅科学研究计划项目(15JK1689)
【分类号】：TP391.41

【相似文献】