基于机器视觉的羊体体尺测量方法研究

Research on sheep body size measurement method based on computer vision

羊体体尺直接反映羊生长发育情况,利用机器视觉实现羊体体尺无接触式测量.首先,采用辅助标识法羊体提取背部轮廓曲线;采用D-P算法和海伦-秦九绍公式寻找曲线曲率最大的点作为臀部测点;其次,分析轮廓曲线特征信息,提取关键帧.最后,在寻找到的关键帧的基础上,寻找肩胛点.结合空间分辨率计算出羊体体尺参数.实验结果表明,可以准确提取关键帧,体尺测量平均误差不超过3%.

The measurement of sheep body size directly reflects the developing situation of the sheep. Machine vision was used to realize sheep body measurements without contacting the sheep body. Firstly,auxiliary identification method was adopted to extract sheep ＇s back contour curve. The maximum point of the curve curvature as a hip measuring point was obtained by using D-P algorithm and Helen-Qin Jiushao formula. Secondly,the contour is analyzed to get its feature information and to extract key frames. Finally,the shoulder point was found on the basis of the found key frames,and the sheep body parameters were calculated combined with spatial resolution method. The experimental results show that the key frames can be extracted accurately and the error for body measurement is less than3% on average.