一种工业无损检测超声图像降噪方法

A Denoising Method for Industrial Nondestructive Detection Ultrasound Images

针对传统的工业无损检测超声图像散斑噪声降噪方法,不能很好的保持图像边缘和细节,提出了一种新的基于粗集与小波的工业超声图像降噪方法。该方法通过考察散斑噪声的统计特征,利用粗集中的等价关系将工业超声图像划分为若干子图,对子图进行对数变换后完成小波变换,采用自适应阈值函数处理小波系数,应用小波逆变换和指数变换得到降噪后的子图,将降噪后的子图进行叠加得到最终的降噪图像。通过新算法在实际工业超声图像中的应用,实现了在保证降噪效果的同时能够较好地保持图像边缘及细节特征,验证了算法的有效性。

To overcome the traditional denoising methods of industrial nondestructive detection ultrasound images could not preserve edges and details well, a new denoising method was proposed for industrial ul- trasound images based on rough set and wavelet. In the method, the industrial ultrasound image was par- titioned into different sub-images based on equivalence relation in rough set, then the logarithm transform for each sub-image was operated and wavelet transform was accomplished, the wavelet coefficients were modified by the adaptive threshold function. The denoised sub-images were obtained by inversing wave- let transform and exponential transform. The final image was obtained by adding the processed sub-ima- ges. Through the new algorithm was applied in the industrial ultrasound images, the results show that the proposed method is more effectively in denoising, preserving edges and details.