基于机器视觉的目标识别分类方法

Target Recognition and Classification Method Based on Machine Vision

工业现场生产的产品批量小、更新快、品种多,存在工作效率低、检测或识别质量得不到保证等问题。为了解决此问题,提出了一种基于机器视觉的目标识别分类方法。采用线阵相机,通过控制产品运动,获取产品表面完整的信息图像,利用多层感知器对图像拼接和处理后的字符进行识别,确定产品型号及批次,引导机器人完成产品的准确分类。实验结果表明,方法识别时间18 ms,识别准确率98%,可有效解决人工作业的各种问题。

The products produced in industry field have the characteristics of small batches,fast renewal,and varieties,where exist some problems such as low efficiency and unguaranteed quality of detection or identification.To solve these problems,a target recognition and classification method based on machine vision is proposed in this paper.The linear array camera is used to obtain the complete information image of the product surface by controlling the product movement.After image stitching and processing,the multilayer perceptron is used to identify the characters on the product,so as to determine the product model and batch.The robot is guided to complete the accurate classification of the products.The experimental results show that the recognition time is 18ms,the recognition accuracy is 98%,which can effectively solve various problems of manual work.