Research Of Livestock Body Size Measurement System Based On Computer Vision

The data of sheep body size is one of the main parameters which grazing sheep industry concerned，also is a key index of sheep breeding and feeding management. Especially in sheep breeding station, body weight, body height, body length, hip width and other body size data are important index for seed selection. In the aspect of scientific breeding，establish a corresponding information database for every sheep could provide an accurate prediction, so as to guide scientific breeding for pasture breeding，and to maximize the industrial economy. However，in the current domestic farms，the traditional measurement techniques of sheep body size has been seriously inconsistent with the needs of our current society. Manual graphometer, caliper, tape and other equipment for manual measurement are still used widely，the workload of recording and processing measurement data relies on workers is very large. In view of the limitation of existing pasture artificial measurement and the need of the development of the modern farming，this paper proposes a construction method of non-contact measurement of sheep based on machine vision.Considering the status of pasture breeding，this research takes on the platform of embedded Linux operating system，acquires the overlooking image of the sheep body as a whole and profile image through the image acquisition system consisting of two wide Angle camera，then calls the real-time computer vision library OpenCV to process the sheep body color images collected，the sheep body contour image is got through the background removing，image denoising, image binarization, contour detection and extraction, then has a body measurements feature point detection and coordinates extraction on contour image，finally the formula of body measurements is used to calculate to get sheep body height，body length，hip width and other body size data. Because this study is a development of the application based on the realistic demand，this system has developed the server application based on multithreading in the local master controller which named BB-Black of the Ubuntu platform. The server program consists of the following sections: the measurement algorithm of sheep body size based on machine vision，the man-machine interface program based on Qt，TCP/IP network communication and serial communication program based on the environment of Linux, electronic scale processing module driver based on STM32microprocessor. Android mobile phone client based on local server and remote data monitoring platform also are developed to realize remote access and control. This system has realized application research and development based on multi sensors data acquisition system，the main controller，the remote data center and the mobile terminal.This system takes on the platform of embedded Linux real-time operating system, combined with the technology of radio frequency identification (RFID)，wireless network communication technology (GPRS/3G) and digital image processing technology (OpenCV)，realizing non-contact measurement and remote monitoring oflarge grazing sheep body weight，body height，body length，hip width growth data. The results of experiment show that：the system is stable，data transmission is reliable. This measurement method has some error，but the average relative error is less than5%，so the application prospects and the actual use value of this system is very high.