基于嵌入式机器视觉的羊体体征测量系统

Measuring System of Sheep Body Size Based on Embedded Machine Vision 

摘要:

针对现有牧场牲畜测量的工作量极大及人工测量的局限性难题,提出了一种基于机器视觉的草原牧场羊体体征的非接触式测量构建方法；考虑到牧场养殖环境,采用在便携式嵌入式Linux实时操作系统平台上,通过广角摄像头获取羊的彩色图像,并调用开源计算机视觉库OpenCV对图像进行实时处理,寻找体征特征点并计算羊的体高、体长等体征数据；图像处理过程:背景差分、图像去噪、二值化、轮廓提取和特征点识别；经实验验证表明:此测量方法有一定误差,但平均相对误差不超过3%,因此该测量方法具有较好的实用价值。

Abstract:

In view of the existing pasture livestock measuring workload greatly and the limitations of manual measurement problem, we propose a non-contact measurement building method of grassland pasture sheep body signs based on machine vision. Considering the pasture breeding environment, this study on the portable real-time embedded Linux operating system platform, through the wide-angle camera to obtain sheep color images, and call the open source computer vision library OpenCV for real-time image processing, looking for signs of feature points and calculate the sheep's body height and length and other signs of data. Image processing:background subtraction, image denoising, binarization, contour extraction and feature point recognition. The experimental verification showed that:this measurement method has a certain error, the average relative error is less than 3%, so this measurement method has great practical value.