微间隙焊缝磁光退化图像自动恢复方法仿真

Simulation of Automatic Recovery Method For Magneto-Optic Degradation Image of Micro-Gap Weld

针对传统微间隙焊缝磁光退化图像自动恢复中,普遍存在着图像恢复信噪比较低、平均绝对误差较大、完成时间较长等问题,提出一种基于离散余弦变换的退化图像自动恢复方法。利用投影追踪学习网络对微间隙焊缝退化图像进行模拟,估计图像退化因子;同时利用小波变换方法对退化图像噪声进行滤波处理。利用平滑图像中具有显著边缘的模具估计成像系统的退化函数(点扩散函数)。采用高斯拟合校正估计所得的图像点扩散函数,将微间隙焊缝磁光退化图像中基本小块合成一个较大的矩阵,获取图像增强系数,根据图像中不一样的波段存在着不一样特点,利用自适应图像增强系数对微间隙焊缝磁光退化图像进行增强,完成图像自动恢复。实验结果表明,所提出方法对退化图像恢复信噪比较高、平均绝对误差较小、完成时间较短。

Traditionally,the automatic restoration for magneto-optic degraded image of micro-gap weld has low signal-to-noise ratio,large mean absolute error and long completion time.In this paper,an automatic restoration method for degraded image based on discrete cosine transform was presented.The projection pursuit learning network was used to simulate the micro-gap weld degraded image and estimate the image degraded factor.At the same time,the wavelet transform method was used to filter degraded image noise.Then,the mold with significant edge in smooth image was used to estimate degradation function(point spread function)of imaging system.In addition,Gaussian fitting correction was used to estimate the point spread function of image.Meanwhile,the basic blocks in micro-gap weld magneto-optical degradation image were combined into a larger matrix to obtain image enhancement coefficients.Because different bands in image had different characteristics,the adaptive image enhancement coefficients could be used to enhance the magneto-optic degradation image of micro-gap weld.Thus,the image could be automatically restored.Simulation results show that the proposed method has higher signal-to-noise ratio for restoration of degraded image,less mean absolute error and shorter completion time.