基于模糊逼近计算视觉算法的奶牛体重测量

The Cows Weight Calculation Based on Non-Measurement Machine Vision

奶牛不同部位的结构较为复杂,不同部位几何特征差异较大,几何特征的曲线特性很强,用直线很难准确对奶牛整体轮廓特征进行逼近描述。传统的接触式测量方法弊端较为明显,造成其乘以体重密度系数后,体重结果误差较大。提出了一种基于模糊逼近计算视觉算法的奶牛体重测量方法。利用平滑滤波方法对机器视觉图像进行去噪处理,对奶牛特征空间位置进行模糊逼近定位,建立奶牛体型立体空间计算奶牛体重。实验结果表明,该算法能够提高奶牛体重测量的精度。

For measured cows weight accurately, the structure in different parts of cows is relatively complex, different parts geometric feature has big differences, geometric characteristics of strong measures, as in a straight line is difficult to accurately on cow overall outline feature .description approximated. The traditional calculation method only can calculate the cows probably outline, its multiplied by weight density number, cause the weight include error. Put forward a new kind of machine vision dairy weight non-contact measurement technology. Use smooth filtering methods to deal with the noise image, the cows in the feature space position, conducting the cow body build stereo space, according to body shape calculation cow weight. Experimental results show that the algorithm can improve the cow weight the accuracy of measurement.