Size Parameter Measurement Of Cow Body Based On Machine Vision

Cow body size parameter is to evaluate the situation and cows grow a very important indicator of breeding cows contain body height, body length, rump angle, teat length and other parameters of dairy farming is important. Traditional cow body measurement data is manually direct contact measurement obtained this workload measurement, measurement conditions are relatively hard, impact stress on the cows. Machine Vision Vision Metrology has observed objects are not in contact, does not appear visual fatigue, high accuracy, stability,and other advantages, the visual measurement techniques used in dairy cattle body size measurement is carried out, the machine vision technology in the field of livestock breeding cows, can be solved to a great extent caused by human shortcomings measure to achieve the refinement of cows fed farming.This topic applies to dairy cow type is detected based on the selected combination of image processing software, the calibration, contour extraction, SIFT feature matching technology, which aims to get the cow's body data to achieve dairy cow type vision measurement. The main contents are as follows:(1) Based on the technical route reference to a number of domestic and foreign literature,describes the vision measurement in the field of dairy cow type measurement applications.(2) The traditional camera calibration is described in detail, and to establish an imaging camera model; introduces the principle of binocular vision measurement, vision measurement and a mathematical model; describes the use of a flat checkerboard calibration object of Zhang Zhengyou calibration method, for the selected software designed to produce circular marker 77? calibration plate by monocular and binocular calibration experiments to obtain internal and external camera parameters of the camera.(3)Choose a powerful extract topic of HALCON image processing software, the outline feature cows for research; and the visual measurement system are introduced, it is the type of camera, lens and image capture cards and other hardware needed for the selection of the measurement system. Research on machine vision technology more commonly used SIFTfeature point detection and matching, detailing the SIFT features matching algorithm steps,first by the image detection multi-scale spatial extreme points, followed by obtaining feature keys and key precise positioning point of the analysis, and then calculate the main direction of the key points, and ultimately to the construction of descriptors to obtain feature vector image feature, wherein the image detection and precise positioning of the key points of the multi-scale spatial extreme point, both the role of extracting extreme point in DOG scale space and optimized in order to gain the feature point.By preliminary tests verified that the measuring system can achieve cows body size measurements, and with job stability, high measurement accuracy, for further study of cow size and body condition, digitized culture foundation, with a very broad application prospects.