Digital image technique and theory of invariant moments were applied to the characteristic extraction of corrosion morphology based on corrosion images containing a large amount of corrosion information. 7 invariant moment characteristics of corrosion image were used to express corrosion morphology. The invariant moments were considered as characteristic parameters and inputs in the probabilistic neural network, and the grade of corrosion was evaluated as output. Accelerated corrosion test in EXCO (exfoliation corrosion) solution was taken as example. The results showed that as a highly concentrated image feature, invariant moment can represent the mapping relationship from corrosion feature to corrosion morphology. This method is simple and has a high recognition accuracy of 87.95%, which can meet engineering requirements.%基于腐蚀图像包含大量腐蚀信息的客观现实,将数字图像技术和不变矩理论应用于腐蚀图像预处理和特征提取,用腐蚀图像的7个不变矩特征值来描述腐蚀形貌特征,利用概率神经网络模式识别技术,建立了以不变矩为特征参数的概率神经网络模式识别模型,实现了金属材料腐蚀等级的评定.以铝合金材料在EXCO溶液中加速腐蚀等级评定为例,分析表明,不变矩作为一种高度浓缩的图像特征,能够表示腐蚀形貌与腐蚀特征的映射关系,该方法简单易行,识别率可达到87.95％,满足工程应用要求.
展开▼
