基于树形小波变换的断口图像识别

Recognition of fracture image based on tree wavelet transform

断口图像的识别是进行失效分析的一个重要组成部分,本文提出一种基于小波的断口识别方法,该方法采用树形小波对金属断口图像进行二级小波变换,剔除近似分量,对其余的各个频带输出的标准差作为断口识别的特征,并采用最小距离分类法进行分类,实验证明,此方法对细节丰富的断口的识别率很高。

Fracture image recognition is a important part of failure analysis.A new method of recognition of fracture image based on wavelet transform is presented in this paper.Two-lever wavelet transform is applied to metal fracture image recognition by selecting tree structure wavelet transform.Standard deviations of each wavelet transform frequency band output except approximate value is used as the characteristics of fracture image recognition.And minimal distance classifier is used to classify the fracture image.Experiment results show that it has a high recognition rate for metal fracture image classification.