基于环投影特征的汽车轮毂定位与识别

为了实现汽车轮毂在辊道上随机混流状态下的型号自动识别,提出一种基于环投影特征的轮毂识别算法。该算法不受轮毂的角度和位置变化的影响。首先提取每种型号轮毂的半径及其环投影特征作为模板,然后求取待匹配图像的重心来近似轮毂的圆心,并在其附近分析每一点的环投影值来确定真正的圆心坐标与半径。利用轮毂的半径和环投影与模板进行粗匹配,筛选掉不匹配的型号。最后通过比较径向投影,返回相似度得分最高的型号。实验结果表明:该算法具有较高的稳定性和实时性,识别速度为1个/2 s。

Positioning and Identification of Automobile Wheel Hub Based on Ring Projection Feature

In order to realize the automatic identification of the model of the car hub under random mixed flow on the roller, a hub recognition algorithm based on ring projection feature was proposed, which was not affected by the angle and position change of hub. Firstly, the radius of each type of hub and its ring projection feature were extracted as templates, then the center of gravity of the matched image was obtained to approximate the hub’s center of gravity, and the real center and radius were determined by analyzing the ring projection value of each point nearby. Rough matching between hub radius and ring projection and template was used to screen out the mismatched models. Finally, by comparing the radial projection, the model with the highest similarity score was returned. The experimental results show that the algorithm has high stability and real-time performance, and the recognition speed is 1/2 s.