Activity Analysis of the Captive Flocks Based on Machine Vision

Abstract. The behavior including attack, movement, foraging and other activities is the external manifestations of health status of chickens, by monitoring and analysis the Chicken activity earlier, it can bring the benefits to find the suspected individual disease and reduce the economic losses. In this paper, the objects of the research are the flock of the chicken, the use of machine vision technology on the flocks about the small movements and the large movement to analyze and identify, proposed to the degree of activity (ratio of the pixel showing the movement in the flock image and the total projected are of the flock) to quantify the movement of flocks, which can determine the health of flocks. Using the self-built visual system to obtain the surveillance images of the flock in real time, the total projected area of the flock can be obtained by taking the mean value after taking the difference between the chicken appearance image and the no chicken appearance image. The pixels in the single image which mean the movement is obtained by the difference of the adjacent image; the optimal stacking time is chosen to superimpose to obtain the pixel points showing the micro-movement in the flock image for a certain period of time; In order to get the large movement of the pixel in chicken flock, the image of optimal time interval selected should be taken the difference; according to the number of the pixels points, the activities of the micro-movement and large movement can be obtained. The result of the research shows that the activity of micro-movement in experimental chickens was stable at 1.5-2, and the activity of large movement was stable in the vicinity of 2, which had some reference significance for the monitoring of flock health.