毕敏娜,张铁民,庄晓霖,焦培荣.基于鸡头特征的病鸡识别方法研究[J].农业机械学报,2018,49(1):51-57.
BI Minna,ZHANG Tiemin,ZHUANG Xiaolin,JIAO Peirong.Recognition Method of Sick Yellow Feather Chicken Based on Head Features[J].Transactions of the Chinese Society for Agricultural Machinery,2018,49(1):51-57.
摘要点击次数: 2322
全文下载次数: 1467
基于鸡头特征的病鸡识别方法研究   [下载全文]
Recognition Method of Sick Yellow Feather Chicken Based on Head Features   [Download Pdf][in English]
投稿时间:2017-02-07  
DOI:10.6041/j.issn.1000-1298.2018.01.006
中文关键词:  机器视觉  病鸡识别  鸡头修正算法  共生特征矩  平均识别算法
基金项目:国家自然科学基金项目(51177053)和广东省省级科技计划项目(2014A050503061)
作者单位
毕敏娜 华南农业大学 
张铁民 华南农业大学 
庄晓霖 华南农业大学 
焦培荣 华南农业大学 
中文摘要:健康黄羽鸡鸡冠鲜红、纹理均匀,鸡眼圆润有神,病鸡鸡冠萎缩变色、鸡眼半闭或全闭。本文提出了一种基于鸡冠及鸡眼构成的鸡头特征信息的病鸡识别方法。首先通过R、G、B分量色差信息去除背景,分析鸡身和鸡冠样本区域H分量分布特点,提取H分量分割阈值分割黄羽鸡。再利用H分量进行阈值分割得到黄羽鸡和鸡冠鸡垂。提出一种利用鸡冠和鸡垂轮廓上两点距离小于阈值的鸡头合并算法,再通过修正算法识别鸡头。在鸡头中,分别提取鸡眼瞳孔轮廓并获取形状几何特征,提取鸡冠的H分量共生矩阵特征,构成基于鸡头特征,采用ARA特征选择算法获得病鸡特征向量,采用支持向量机(SVM)分类器进行训练分类。实验结果表明,病鸡识别正确率为92.5%,表明利用机器视觉获取鸡头特征进行病鸡识别具有可行性和研究价值。
BI Minna  ZHANG Tiemin  ZHUANG Xiaolin  JIAO Peirong
South China Agricultural University,South China Agricultural University,South China Agricultural University and South China Agricultural University
Key Words:machine vision  sick chicken recognition  head correcting algorithm  co-occurrence matrix  average recognition algorithm
Abstract:Yellow feather chicken is a kind of traditional poultry in China. In recent years, labor shortages and diseases are two factors that affect production and health of yellow feather chicken. Using machine vision instead of manual monitoring of chickens and rapid identification of diseased chickens is a new research direction to solve the plight of poultry breeding. The symptoms of a diseased chicken mainly occurred in the eyes and comb which were mainly in the head. A method to recognize diseased chickens based on the head features was proposed. At first, totally 500 images of the healthy chickens and 236 images of the diseased ones were separately captured from the videos taken in the actual feeding environment composed the graphic base for the experiment. From each image, 10 chicken areas with 10 pixels×10 pixels, three comb areas with 5 pixels×5 pixels and 10 background areas with 10 pixels×10 pixels were selected to form the sample sets. The R,G,B and H component values of each pixel in the sample sets were obtained to get color difference threshold of the background and H threshold of the comb and chicken. Segmentation of the yellow feather chicken could be completed by using the color difference threshold for removing the background and the H thresholds for the object regions. Combs and eyes and their contours were also obtained in HSV color model by the H threshold. Then distance d1 between any two points on the different contours of the combs was calculated. The points if d1 was less than the maximal side of the external rectangle of each comb were connected to make the two combs to be a whole. The external rectangle of the connected combs included the chicken head. A correcting method was to extend this external rectangle to both sides until the width of the rectangle was larger than the height to improve the integrity of the chicken head recognition. Co-occurrence matrix of H component was calculated from the comb to get the color feature and the texture feature including ASM,COR,IDM,Ent 6 geometry features including A,P,R,E,C and A/P were computed from the eye contour. All these 10 features formed characteristic vector of the chicken head. Average recognition accuracy (ARA) was adopted to select the best feature set (ASM,Ent,E and A/P). At last, SVM was trained by 420 healthy images and 200 diseased images, and then tested by 80 healthy chicken images and 36 diseased chicken images. Accuracy of the identification of diseased ones was 92.5%. The test results indicated that the proposed method was feasible and valuable to identify diseased chickens by head features using machine vision.

Transactions of the Chinese Society for Agriculture Machinery (CSAM), in charged of China Association for Science and Technology (CAST), sponsored by CSAM and Chinese Academy of Agricultural Mechanization Science(CAAMS), started publication in 1957. It is the earliest interdisciplinary journal in Chinese which combines agricultural and engineering. It always closely grasps the development direction of agriculture engineering disciplines and the published papers represent the highest academic level of agriculture engineering in China. Currently, nearly 8,000 papers have been already published. There are around 3,000 papers contributed to the journal each year, but only around 600 of them will be accepted. Transactions of CSAM focuses on a wide range of agricultural machinery, irrigation, electronics, robotics, agro-products engineering, biological energy, agricultural structures and environment and more. Subjects in Transactions of the CSAM have been embodied by many internationally well-known index systems, such as: EI Compendex, CA, CSA, etc.

   下载PDF阅读器