关键词: 图像处理, 鸡冠测定, 样本图像, Kmeans聚类

Abstract: Chicken comb is the most obvious and important feature of chicken secondary sexual characteristic. Determination of area and color grade of chicken comb has important reference index and auxiliary judgment effect for chicken scientific research and production. In order to accurately obtain the authenticity of the comb, this study designed a method of area and color level determination by using the comb image processing and feature extraction. 500 healthy 90dayold Shaobo chickens (half of male and female) were used as samples, the comb images were collected for analysis and the color feature data were obtained. Seven cluster centers were extracted by Kmeans clustering as color grade indicators. A special measuring board with a reference pattern was made and the measuring image was distortion corrected and color corrected. The area and color grade of the comb were obtained by the contour discrete set of points pixel area algorithm and color difference distance algorithm. Determination results were verified by digital plane imaging simulation validation method as follows: the relative error range of area measurement value and set value was among 0.11%-4.84%, the average relative error was 2.24%, and the color level accuracy was 96.67%. The results showed that this method had a high accuracy. This research provides a new measurement method for rapid and effective acquisition of comb trait parameters.

Key words: image processing, comb determination, sample image, Kmeans clustering