Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Abstract. Now, artificial inspection is the only way to distinguish the health behavior of caged poultry, which is physical energy consuming, poor consecutiveness of observation and harmful to people. In order to achieve the automatic monitoring of dead or sick poultry, a method of health behavior recognition based on computer vision is proposed in this study. The first step to identifying the sick chicken in the cages by machine vision system is segment of object from images fast and correctly. However, it is a challenge to extract the chicken from pictures because of the complex background. A segmentation method based on the difference of three components of RGB model m was presented to extract hens‘ leg from the image. Totally 20 images were selected from 100 images to setup pixels data sets for the color components analysis. All data sets were analyzed in the different color models, such as RGB and HSV. It was found that the value of R, G, B components of the background and the chicken was nearly the same or very close. The difference between the R, G and B components of the chicken leg sample set is obviously greater than the background sample set. This characteristic was used to abandon the background pixels in the RGB model. After the chicken was extracted from image, an algorithm was developed to count the number of hens in cage.