Algorithm of sheep body dimension measurement and its applications based on image analysis

Body size parameters of sheep can reflect its growth development, production performance and genetic characteristics. Therefore, the body size measurement is great significance in sheep breeding. In this study, a measuring method based on vision image analysis for the body size of sheep is proposed, which can be practically applied in farm environment without disturbing the animals. This approach is based on computer-assisted visual image capture in a position-limit apparatus, and based on an automatic foreground area extraction algorithm known as simple linear iterative clustering (SLIC) SuperPixels and Fuzzy c-means (FCM) clustering, a center line of flexible symmetrical body extraction algorithms, as well as measuring points extraction algorithms. The test results for 27 small-tailed Han sheep chosen randomly from herd show that the method for foreground extraction can get segmentation images with well-remained boundaries. Detection algorithm of the center line of sheep body in top view has a relatively high adaptability. The extraction of measurement points in different postures for sheep's body size has a better stability and accuracy, the maximum average relative errors between the detected and measured values of body height, rump height, body length, chest depth, chest width and rump width are 1.13%, 1.54%, 2.03%, 4.45%, 2.25% and 2.41%, respectively. The use of both left view and right view can improve the precision of the measurement, and the values from one test may have a greater deviation from the actual values due to the variety of sheep body posture, but the accuracy can be improved by averaging the measurements repeated for many times. The results also show that measurement of sheep size based on vision image analysis is feasible, and it can ensure accuracy, reduce workload and sheep stress compared to the method conducted by man. Prediction result of live-sheep weight based on body size shown that the parameters got by image processing can be used for monitoring the growth of sheep.