摘要
针对传统的灰度值模板匹配算法计算时间复杂度高的问题,结合序贯相似性检测算法与图像金字塔搜索策略,设计一种基于灰度的快速模板匹配方法。先对图像进行直方图均衡化,增大图像对比度,通过高斯滤波进行降噪,减少匹配时噪点的干扰,利用阈值分割、边缘检测初步区分出特征区域,减少匹配搜索面积,加快匹配速度。利用序贯相似性检测算法进行度量匹配,结合图像金字塔分层搜索策略,大大加快匹配速度。创建多角度模板,解决大角度偏差无法匹配或匹配概率低的问题。实验证明该方法满足工业应用中系统精度、实时性要求。
Aiming at the high time complexity of the traditional gray-valued template matching algorithm, a sequential gray-based fast template matching method is designed based on sequential similarity detection algorithm and image pyramid search strategy. The histogram equalization of the image is performed to increase the contrast. Gaussian filtering is used to reduce the noise in the image, which reduces the noise during the matching. The threshold region segmentation and edge detection are used to distinguish the feature regions and reduce the matching. Search area to speed up matching. Sequential similarity detection algorithm is used for metric matching, combined with hierarchical pyramid search strategy to further speed up matching, and at the same time, multi-angle templates are created to solve the problem that large-angle deviations cannot match or the matching probability is low. Experiments show that this method satisfies the system accuracy and real-time requirements in industrial applications.
作者
郑剑斌
郑力新
朱建清
ZHENG Jian-bin;ZHENG Li-xin;ZHU Jian-qing(College of Engineering of Huaqiao University,Quanzhou 362021;Industrial Intelligent TechnologT and System Fujian Engineering Research Center of Universities,Quanzhou 362021)
基金
国家自然科学基金青年科学基金(No.61602191)
厦门市科技计划项目(No.3502Z20173045)
关键词
模板匹配
序贯相似性检测
图像金字塔
高斯滤波
Template Matching
Sequential similarity detection
Image Pyramid
Gaussian filter
