基于空间相关性的成捆钢筋图像计数算法研究  

Research on Counting Algorithms of Bundled Steel Bar Image Based on Spatial-correlation

作　　者：袁中锦 刘宇[1] 时娟 YUAN Zhong-jin;LIU Yu;SHI Juan(College of Software, Yunnan University, Kunming, 650221, China)

机构地区：[1]云南大学软件学院

出　　处：《电脑与信息技术》2019年第4期10-13,共4页Computer and Information Technology

摘　　要：为解决钢筋生产企业和钢筋采购部门存在的成捆钢筋计数难题,提出了一种利用空间相关性的钢筋图像处理计数算法。针对钢筋图片中通常存在钢筋较多粘连且对比度较低的问题,提出直方图均衡化和模板滤波增强相结合的算法,算法主要分为以下几步:首先,利用直方图均衡化提高采集图像的对比度,接着,应用二维高斯模板对经过对比度增强后的图片进行空间相关性计算,得到特征增强后的钢筋横截面相关性矩阵结果图,最后,对该图像做局部极大值计数检测得到最终计数结果。经实验证明,此计数算法在实际环境下检测准确度能达到98%,符合应用需求。To solve the problem of counting bundles of steel bars in steel bar production manufacture and steel bar procurement departments, an image processing algorithm of steel bar counting based on spatial correlation is proposed. Aiming at solving the reinforcing bars adhesion and the low image contrast of reinforcing bars problem, an algorithm combining histogram equalization and template filtering enhancement is proposed. The algorithm is mainly composed of three steps: Firstly, histogram equalization is used to improve the contrast of the collected images;In addition, two-dimensional Gauss template is used to calculate the spatial correlation of the images after contrast enhancement. Finally, the final counting result is obtained by counting the local maxima in the image. Experiments show that the detection accuracy of this counting algorithm can reach 98% in the actual environment, which meets the application requirements.

关 键 词：成捆钢筋计数 数字图像处理 直方图均衡化 空间相关 高斯模板 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]