摘要
小波阈值法在图像去噪中应用较广泛,该方法最重要的一个环节是最优阈值的确定.为此,提出一种新的自适应多阈值的阈值计算方法.由于小波分解后,信号小波系数的绝对值较大,噪声小波系数的绝对值较小,并且不同尺度不同方向上噪声的方差不同,方差和信号小波系数的个数存在一定的关系,这样就可以根据信号小波系数的个数确定最佳阈值在小波系数绝对值序列中的位置,得出最佳阈值.实验表明,使用本方法从RMSE和SNR两个客观指标上看,能得到更好的效果,同时更适合人眼的视觉特性.
Wavelet threshold is widely used for image denoising, one of the most important parts of the method is calculating the optimum thresholds. So, in this paper, a new method of adaptive multi-threshold for calculating optimum thresholds is proposed. After wavelet decomposition, the absolute values of signal wavelet coefficients are larger, the absolute values of noise wavelet coefficients are less, and the noise variances are different based on different scales and orientations, certain relation exists between variances and the number of signal wavelets, so the positions of optimum thresholds can be obtained in the sequence of wavelet coefficient absolute values according to the number of signal wavelet coefficient. Experiments show that the better effects can be obtained through the new method according to the objective indexes RMSE and SNR , at the same time, the results are of more adaptive vision character of human eyes.
出处
《沈阳理工大学学报》
CAS
2008年第4期10-12,58,共4页
Journal of Shenyang Ligong University
关键词
小波变换
图像去噪
阈值
wavelet transform
image denoising
threshold
