Abstract: According to the theory of machine vision and ANN, a pattern recognition method for online prediction of sintering quality is put forward. Based on some sintering plant, the vision characteristics which affect sintering quality are analyzed. And the images of sintering at the end of the sintering machine are processed. Images’ spatial lower order moments are used to describe those vision characteristics. As a result, most effective characteristics for classification and recognition are selected as the input of ANN. An improved BP classifier is proposed to realize sintering quality online prediction as well. Experimental result shows that the proposed method is feasible.

Key words: vision characteristic, image processing, sintering, ANN

摘要： 应用机器视觉和人工神经网络理论提出了对烧结质量在线判断的一种模式识别方法。以某烧结厂为研究背景，分析影响烧结质量的视觉特征，从烧结机机尾摄取断面图像并进行处理，用图像的空间低阶矩描述目标的视觉特征，从而可以选出对分类识别最有效的特征作为人工神经网络的输入，构造改进的BP神经网络分类器，实现在线判断烧结质量，实验证明了该方法有效可行。

关键词: 视觉特征, 图像处理, 烧结, 人工神经网络

CLC Number: 

• TP273