Development of machine vision system to detect meat mass of pig prior to slaughtering
暂无分享,去创建一个
Abstract. Pig breeding has become one of the promising agricultural businesses. Huge consumption of pork in home and abroad has attracted breeders into pig breeding business. The meat mass from individual pig after slaughter determines the economic benefit. Prior-slaughter of pig before maturity can bring huge loss to the meat industry. This study focused on development of machine vision system to predict the meat mass of pig prior-slaughtering. A hardware system with floor area of 1m x 1.8m (with inbuilt weighing machine) equipped with two cameras, was designed to acquire image of pig. The cameras were mounted to acquire top view and side view images of pig. Image processing algorithm was written in C++ to obtain the body dimensions of the pig, namely the body height, body length, chest width and hip width using the acquired images. The original top view and side view images were converted into grayscale and binary image. Region of Interest (ROI) was obtained from the binary image and edges were detected. Algorithm was developed to measure the body dimension based on the edge points in the image. The body length, chest width and hip width were obtained from the top view image whereas the side view image was used to calculate the body height. Compared to actual body dimensions of 21 pigs, the results obtained by image processing algorithm have high accuracy with maximum error of less than 3%. In order to verify that the pig body dimensions with weight were better at detecting the meet mass than weight only, a data analysis using multiple linear regression (MLR) method was carried out. Result showed that the former method to predict meat mass was better. The system being cost, labor and time efficient, can be used for detection of meat mass of pig in commercial pig farming.