Research And Implementation Of Yak Body Size Measurement System Based On Machine Vision

The body size and weight parameters of yaks are important indicators for monitoring the growth and health of yaks,and have important reference value for the fattening,reproduction and selection of breeding cattle.In Tibetan areas of China,the body size parameters of yaks are usually obtained by herdsmen using traditional measurement tools for manual contact measurement,but the manual measurement method is easy to stimulate yaks,with certain risks,and there are problems of heavy workload and low efficiency.In view of the above background and problems,this thesis designs and implements a non-contact yak body size and body weight measurement system based on machine vision.When the yak passes through the measurement device,it automatically detects the position of the yak and measures the weight and body size,judges the health status of the yak according to the measurement results,and realizes group feeding.The main work of this thesis includes the following parts:1.Yak position detection based on Dense Inverse Search-based method.Compared with several commonly used moving target detection algorithms,DIS algorithm which has strong adaptability to device vibration and can detect the complete yak trunk is selected for yak position detection.Calculate the dense optical flow field in the region of interest,determine the motion state and current position of yak according to the change of optical flow vectors,and judge whether yak completely enters the measuring device.2.Yak contour segmentation based on image processing algorithm.After the yak completely enters the measuring device,use the camera to collect the side view and top view of the yak.For the side view of yak,combine Hough transform and morphological method to detect the straight line where the railing is located;Make a mask to remove the foreground railing;The image inpainting algorithm combined with FMM algorithm and Criminisi algorithm is used to inpaint the side view of yak after removing the shielding railing;The binary image of yak side view is segmented by Otsu method.For yak top view,on the basis of removing noise by combining HSV color space and image difference method,the threshold segmentation method is used to obtain the binary image of yak top view.3.Yak body size calculation based on feature point extraction.Extract the yak edge contour from the binary image.The envelope method is used to calculate the coordinates of trunk line segment,deviation correction angle and body width feature points on the back contour of yak;It is calculated that the maximum curvature point on the hip contour in the side view of yak ’is the sciatic point,and the highest point on the back contour in the side view of yak is the bun point.Calibrate the camera and calculate the pixel scale coefficient on the calibration plane,correct the position and posture of yak by using the deviation correction angle,deduce the pixel scale coefficient of the plane where the trunk straight line is located after deviation correction by using perspective transformation,and calculate the parameters such as body height,body width and body length in Yak’s body size in combination with the scale coefficient.The experimental results show that this method can effectively measure the body size parameters of yak,and the relative error of the measurement results is small,which is within the allowable range of body size measurement error.Compared with the traditional manual measurement method,the measurement system studied in this thesis is more efficient and safer,has high application prospect and value,and provides a new way for the measurement of yak body size parameters.