基于快速模板匹配的金标试纸条图像配准

Inproved Template Matching Algorithm for Processing Test Strip Image Abstract

随着金标试纸条在医疗、水质和食品等众多领域的广泛使用,以及国家对各类检测标准要求的提高,研制可以对金标试纸进行定量检测的仪器具有重要的实际意义.考虑到机器视觉在工业界的广泛应用,可以通过工业摄像头获取金标试纸条图像,利用合适的图像处理技术,提取相应的特征,从而可以实现对金标试纸条的定量检测.为了解决工业相机采集到的金标检测卡图像存在的图像倾斜和试纸条区域寻找问题,在前人研究的基础上,根据金标检测卡外形的特点采用基于LSD的Hough变换来进行图像校正,并提出了一种改进的基于水平投影直方图的模板匹配算法.与传统的模板匹配算法相比,该算法在保证匹配正确率的前提下,具有更快的匹配速度,并且随着待匹配图像的增大,其所耗费的匹配时间的增长率较小.

It is practically important to develop an instrument for quantitative testing strips as the strips were widely used in areas like health, water, food, etc. and national higher demands were made on testing standard. Considering the widely use of machine vision, the test strip image can be got from the industrial camera. Taking advantage of the image processing for extracting corresponding feature can help achieve the quantitative test of the strips. In order to overcome the difficulties that rectificating the sloping test strip images and locating the region about test strip, this paper adopts the hough transform to correct the test strip image and proposes an improved template matching method based on the shape of integral projection. Compared with the traditional methods of template matching, this new method has much faster matching speed with guaranting the high matching accuracy. Moreover, the growing time of method proposed in this paper increase slowly along with enlarging the test strip image.