Online detection of abnormal chicken manure based on machine vision

Abstract. The excretion status of livestock and poultry is one of the important indicators to monitor their health status. At present, in the stacked chicken coop, the inspectors make preliminary judgment on the health status of chickens by observing the color and trait of chicken feces on the manure belt daily. However, this monitoring method is highly subjective and disturbing to the chickens. In this paper, we propose a chicken manure image recognition method based on machine vision online monitoring to make preliminary judgments on chicken health status. The method firstly preprocesses the online collected chicken manure images, and makes preliminary judgment on abnormal chicken manure by color extraction algorithm, extraction of abnormal area contour detection, calculation of area ratio and comparison with pre-set threshold; then adopts module matching algorithm and analyzes and compares the grayscale characteristics of the sample and the image to be tested to make further judgment on whether the chicken manure is normal or not. The test shows that the method is effective for online monitoring of images with abnormal chicken droppings, and can initially determine the health condition of chickens.