基于PGH矩改进SURF算法的工业散料识别研究  

Research on Industrial Bulk Material Identification Based on PGH Moment Improved SURF Algorithm

作　　者：王萌萌[1] 李春明[1] 刘海洋[1] 杨鹏飞 武少广[1] 

机构地区：[1]河北科技大学信息科学与工程学院,河北石家庄050000 [2]石家庄市京华电子实业有限公司工程部,河北石家庄050000

出　　处：《数学的实践与认识》2017年第13期143-149,共7页Mathematics in Practice and Theory

摘　　要：针对工业散料识别过程中图像特征匹配率低的问题,提出一种基于PGH矩的改进SURF图像匹配算法.首先,研究Gaussian-Hermite矩,将其扩展到复数空间中,推导出Polar-Gaussian-Hermite矩;其次,利用升降算符法计算图像PolarGaussian-Hermite矩,获得新的特征向量;最后,将原始图像提取的特征点和模板图像进行准确匹配得到最优匹配结果.实验结果表明算法能够解决缩放、旋转和曝光情况下工件匹配问题,误匹配率8%左右,满足工业散料识别系统的准确性和实时性要求.Aiming at the problem of low matching rate of image feature in the process of industrial bulk material identification, an improved SURF image matching algorithm based on PGH moments is proposed. First of all, the Gaussian-Hermite moment is researched and extended to the complex space, the Polar-Gaussian-Hermite moment is deduced in the complex space; Secondly, the method to calculate the image Polar-Gaussian-Hermite moment utilizing raising and lowering operators is presented, and the new feature vector is obtained; Finally, the feature points are extracted from the original image to be metched, which are metched with the template image accurately. The experimental results prove that this algorithm can solve the matching problem of workpiece in the case of scale, rotation and exposure. The error matching rate is about 8%. This algorithm meets the accuracy and real-time requirements of industrial bulk material identification system.

关 键 词：特征匹配 改进SURF算法 GAUSSIAN-HERMITE矩 Polar—Gaussian-Hermite矩 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]