基于Zernike矩及支持向量机的猪的姿态识别

Gesture recognition of pig based on Zernike moments and support vector machines

为监控猪的行为,提出了基于Zernike矩及支持向量机的猪的行走姿态识别方法。首先对原始图像进行预处理,提取出原始图像中猪的轮廓图像。然后,根据标准矩对上述图像进行归一化,再对归一化后的二值轮廓图提取Zernike矩特征。在此基础上,利用支持向量机理论设计了多种姿态分类器,实现对猪的正常行走、低头行走、抬头行走、躺卧等四种姿态进行识别。实验结果表明,此方法对猪的姿态分类识别的准确度达到了95%以上。该项研究对猪的姿态识别方面具有显著价值。

In order to monitor the behavior of the pig, this paper proposes the gesture recognition of pig based on Zernike moments and support vector machines. First, the original image conducts pre- processing, extraction contour image of the original image in pigs. Then, according to the standard moments of the image is normalized, and then the binary contour map normalized Zernike moment feature extraction. On this basis, it designs of a variety of gestures theory classifier by support vector machine to achieve, the pig normal walking, walking down, looked up walking, lying and other four gesture recognition. The experimental results show that the accuracy of the classification and identification of swine gesture reached more than 95 %. The study has significant value for gesture recognition pigs.