基于SVM的图像识别在零件分拣系统中的应用

Application of image recognition based on SVM in part sorting system

针对外观形状、质心分布较均匀的零件在自动化生产中难以实现有方位要求的自动送料问题,对零件的图像采集、图像处理、图像特征提取和分类识别等作了研究和分析。提出了以图像识别为核心的零件分拣系统,在振动盘出料口安装摄像头进行图像采集,对采集的图片进行小波变换,滤去干扰噪声和降维,用主成分分析法(PCA)对图片进一步降维和特征提取,提取得到的特征向量作为支持向量机(SVM)的输入向量,通过SVM对输入向量进行了样本识别,从而确定零件的位置状态,最后通过驱动装置将不符送料要求状态的零件推出送料轨道,从而为下道工序实现自动送料。实验结果表明:训练样本达到20以上时,该方案两种零件的识别正确率都达到了100%,单个零件的识别时间在1 s以内,能满足实际生产要求。

Aiming at solving the problem that the parts with the uniform appearance shape and mass distribution were hard to realize automatic material feeder under the directional requirement,the analysis and selection of the image acquisition,image processing,image feature ex- traction and pattern recognition to the parts were made,and the parts sorting system based on the image recognition was proposed.Wavelet transformwas used on the images that were acquired by installing cameras on the exit of the vibrating disk,and then the disturbance filtering and dimensionality were performed.Principal component analysis (PCA)was further applied to reduce dimensions and extract the features of images,and support vector machine (SVM)was used on the input vectors for pattern recognition.The location of the parts was determined by carrying out SVM on the input vectors,and meanwhile the parts that do not comply with the requirement were propelled by the drive set, for the purpose of realizing automatic feeding in the next procedure.The experimental results indicate that when the number of training samples reaches 20,the recognition accuracies of the two kinds of parts are 100%,and the recognition time of one single part is within one second,which can meet the actual production requirements.