Abstract: Aiming at the questions of traditional method in pattern recognition of surface defect images of cold steel strip, a pattern recognition method based on wavelet transform and genetic algorithms was proposed. Experimental investigations were carried out on computer aiming at four typical surface defect images of cold steel strip, the results showed that the method can recognize correctly the surface defect images of cold steel strip. And, the research on relation between the rate of correct recognition and the number of training samples was carried out.

Key words: cold steel strip, surface defect, wavelet transform, genetic algorithms, pattern recognition

摘要： 针对传统的方法在冷轧带钢表面缺陷图像模式识别中存在的问题，提出了基于小波变换和遗传算法的模式识别方法。结合国内实例，对四种典型冷轧带钢表面缺陷图像进行了实验研究，实验结果表明，该识别方法能够对冷轧带钢表面缺陷图像进行有效识别。另外，还对识别率与训练样本数量的关系进行了研究。

关键词: 冷轧带钢, 表面缺陷, 小波变换, 遗传算法, 模式识别