CAJ | 학술논문

提出了一种基于改进的噪声标准差估计的轮廓波去噪算法，在常用的轮廓波去噪算法基础上提出了新的解决方案。该方案将滤波法与改进的图像结构特征度度量分析算法结合起来，筛选出适合计算噪声标准差的图像子块集合，再用直方图法估计图像噪声标准差，然后将该标准差用于轮廓波去噪。在标准差估计对比试验中，将滤波法、分块法、改进的分块法与文中的标准差估计算法进行对比；在去噪对比试验中，采用基本的小波阈值去噪算法（ universal 阈值），由小波阈值法引申出的普通轮廓波阈值去噪算法，基于维纳滤波的轮廓波去噪算法，基于系数建模的轮廓波去噪算法与文中算法做对比。实验结果表明：文中算法能够更加精确地估计图像噪声标准差，且去噪效果与普通轮廓波去噪及其他轮廓波去噪算法相比更加稳定，鲁棒性更好。
제출료일충기우개진적조성표준차고계적륜곽파거조산법，재상용적륜곽파거조산법기출상제출료신적해결방안。해방안장려파법여개진적도상결구특정도도량분석산법결합기래，사선출괄합계산조성표준차적도상자괴집합，재용직방도법고계도상조성표준차，연후장해표준차용우륜곽파거조。재표준차고계대비시험중，장려파법、분괴법、개진적분괴법여문중적표준차고계산법진행대비；재거조대비시험중，채용기본적소파역치거조산법（ universal 역치），유소파역치법인신출적보통륜곽파역치거조산법，기우유납려파적륜곽파거조산법，기우계수건모적륜곽파거조산법여문중산법주대비。실험결과표명：문중산법능구경가정학지고계도상조성표준차，차거조효과여보통륜곽파거조급기타륜곽파거조산법상비경가은정，로봉성경호。
In this paper, an algorithm based on the modified noise standard deviation estimation was proposed. The proposed algorithm provides a new approach for the foundation of common contourlet denoising methods. The combi?nation of modified image structural characteristic measurement analysis method and filter method is used in selecting the image sub?block, which is suitable for computing standard deviation of noise. Finally, the histograms of those sub?images are used to estimate standard deviation of the image noise, which is subsequently used for denoising of contourlet. In the contrast experiment of standard deviation estimation the filter methods, partition methods and im?proved partition methods are chosen to compare with the standard deviation estimation algorithm. In the contrast ex?periment of denoising, the universal wavelet threshold denoising, the common contourlet threshold denoising ex?tended from wavelet threshold, the contourlet denoising based on wiener filtering, and the contourlet denoising based on coefficient models are chosen to compare with the denoising algorithm proposed in this paper. The experi?mental results showed that the method can estimate standard deviation of images more accurately and performs more reliable than other contourlet denoising algorithms.