一种改进的钢板表面缺陷图像增强算法

An improved image enhancement algorithm for steel plate surface defects

针对钢板表面缺陷图像存在的光照不足、光照不均和对比度低等问题,提出了一种新的图像增强算法。首先利用小波变换将输入的缺陷图像分解成高、低频两部分分量;其次使用基于引导滤波的Retinex算法对低频部分进行增强;同时使用软、硬阈值去噪算法对高频部分进行增强;然后利用小波逆变换进行图像重构,并对重构的图像进行伽马变换。最后,对三种缺陷图像进行测试,其信息熵分别为7.08、7.33和7.57,峰值信噪比分别为47.25、36.33和48.58,图片对比度分别为127.91、134.32和166.03。实验结果表明,该算法具有很好的图像增强效果。

Aiming at the problems of insufficient illumination,uneven illumination and low contrast on the surface defective image of steel plate,a new image enhancement algorithm is proposed.Firstly,the wavelet transform is used to decompose the input defect image into two parts of high and low frequency.Secondly,the low-frequency part is enhanced using the Retinex algorithm based on guided filtering;At the same time,the soft and hard threshold denoising algorithms are used to enhance the high-frequency parts;The image reconstruction is then performed using the wavelet inverse transformation,and the gamma transformation is performed on the reconstructed image.Finally,the information entropy of the three defect images was tested,the information entropy was 7.08,7.33 and 7.57,the peak signal-to-noise ratio was 47.25,36.33 and 48.58,and the picture contrast ratio was 127.91,134.32 and 166.03,respectively.Experimental results show that the algorithm has a good image enhancement effect.