摘要
针对识别图像中数字的需求,本文基于OpenMV实现了一种轻量化的识别方法。本方法首先创建目标数字的字模,然后提取实际图像中数字所在的区域并与各数字字模按像素作差取绝对值得到像素差值,最后确定像素差值最小的字模对应的数字为匹配数字。该数字识别方法占用的硬件资源较少,速度较快,是一种轻量化的机器视觉方法,可以较好地在OpenMV这类小型低功耗、硬件资源有限的机器视觉平台上运行。
Based on OpenMV,a lightweight method to recognize numbers in images is implemented.First,the method creates templates for target numbers,then extracts the region of number in actual image and obtains the pixel difference value by calculating the absolute value of the pixel-wise subtraction of number region image and templates.Finally,the number corresponding to the template with the smallest pixel difference value is the matching number.This number recognition method consumes less hardware resources and runs fast,which is a lightweight machine vision method and can run well on compact low-power machine vision platforms which have limited hardware resources such as OpenMV.
作者
薄天成
BO Tian-cheng(College of Electronic and Information Engineering,Shandong University of Science and Technology,Qingdao 266590,China)
出处
《山东工业技术》
2022年第3期113-117,共5页
Journal of Shandong Industrial Technology
