基于联合模型匹配的齿轮视觉检测方法

针对工业场景下对图像中机械零件的视觉检测任务引入区域块循环分解行人检测算法。根据检测精度的需要，利用Par-King图像增强算法改进目标和背景梯度方向直方图特征的可分性，自建数据集训练出SVM和频域SVR两个不同的分类器模型并利用可变性部件模型算法中图像金字塔匹配方法对测试集图像进行联合视觉检测。实验结果表明，该联合检测方法相比于区域块循环分解行人检测算法有着更好的检测精度。

Method of gear vision detection based on joint model matching

Aiming at the visual inspection task of the mechanical parts in the image under the industrial scene, the circulant pedestrian detection algorithm is introduced. According to the need for detection accuracy, it uses Par-King image enhancement algorithm to improve the target and the separability of histogram features in background gradient direction to train the two different classifier models of SVM and frequency domain SVR through self-built data set. And the visual inspection of the test set image is carried out by using the image pyramid matching method in the variability part model algorithm. The experimental results show that the combined detection method has better detection accuracy to the regional block cyclic decomposition pedestrian detection algorithm.