别克副车架焊缝图像识别算法分析-analysis of seam image recognition algorithm for buick subframe.docxVIP

解,以实现滤波和数据的缩减；然后对分解后的子图像进行以 K-L 变换为基础的 PCA 分析；计算 Euclidian 值后进行识别判断，最后用满足 条件的待测图像更新图像集。多次试验表明，该方法具有良好的容错 性和检测速度。其中 PCA 分析结合了焊缝图像识别的具体实践。其依 据为焊缝形成时的瞬时高温打乱了工件原来比较规则的纹理，形成了 相当明显的不规则性，从而造成已焊图像和未焊图像在主要特征上的 差别。焊缝识别算法在 WINDOWS 操作系统上，在 VC6.0 开发环境下， 运用 Intel 提供的 Open CV 图像库实现。进行了大量的试验，试验结果 证明该焊缝图像识别算法有相当高的准确率，其中 10 条焊缝的判断准 确率达到了 99.17％。关键词：图像识别，小波变换， PCA，K-L 变换, 焊缝ALGORITHM RESEARCH ON IMAGE RECOGNITION ABOUT WELDING SEAM OFBUICK AUTOMOBILE AUXILIARY FRAMEABSTRACTWith the rapid development of modern industry such as metallurgy, machinery，petrifaction, the atom, spaceflight and so on, the weldingindustry becomes more and more important. It has become the third critical kind of fabrication industry, only less important than assembly and machinery process, so the assurance of the welding quality is of big importance. The quality of the welding seam includes shape, position and fault. Especially in the auto industry, it is important to detect and estimate the welding seam.The content of this thesis is one of the key technologies in the Welding Seam Recognition System, which is a project of Shanghai Auto Fund. We developed this system to detect the quality of the 21 welding seams in Buick automobile auxiliary frame. Because the position and the shape of the 21 welding seams are different, a number of CCD cameras are used to get images. All these cameras communicate with computer via USB interface. Then, we process the captured images and estimate the quality of Welding seam.Welding seam image of Buick automobile auxiliary frame is very unstable.Because it is welded by human, and it is got by CCD. The detect module based on median filtering and KL transform is adopted to detect welding seam image. But the question of accuracy and speed is still not sloved.This thesis focuses on the research into the question of accuracy andspeed. In the paper, a new algorithm has been established to solve the speed and accuracy problems in detection of welding seams. Wavelet transform is adopte