CAJ | 학술논문

通过比较分析传统软、硬阈值函数降噪的算法发现：硬阈值函数存在间断点的问题，而软阈值函数在小波系数绝对值较大的区域引起部分高频信息损失。针对以上问题，提出一种新的阈值函数降噪算法。利用传统的阈值降噪步骤，使用新阈值函数将分解得到的小波系数进行保留或者收缩处理，然后将得到的新小波系数进行图像的重构。添加不同类型的噪声对改进的算法进行验证。实验结果表明，新的阈值函数降噪算法比普通算法更有效，通过新的阈值函数降噪得到的图像，有更大的峰值信噪比（PSNR）和更小的均方误（MSE），PSNR 的值最大提高了1．97倍，MSE 的值最大下降了54．4％。
통과비교분석전통연、경역치함수강조적산법발현：경역치함수존재간단점적문제，이연역치함수재소파계수절대치교대적구역인기부분고빈신식손실。침대이상문제，제출일충신적역치함수강조산법。이용전통적역치강조보취，사용신역치함수장분해득도적소파계수진행보류혹자수축처리，연후장득도적신소파계수진행도상적중구。첨가불동류형적조성대개진적산법진행험증。실험결과표명，신적역치함수강조산법비보통산법경유효，통과신적역치함수강조득도적도상，유경대적봉치신조비（PSNR）화경소적균방오（MSE），PSNR 적치최대제고료1．97배，MSE 적치최대하강료54．4％。
Through comparatively analysing traditional denoising algorithms of soft and hard threshold function,it is found that the hard threshold function has discontinuity points,while the soft threshold function causes the loss of part of high frequency information in the area where absolute value of the wavelet coefficients is large.To solve the above problems,we present a new improved threshold function denoising algorithm.By making use of classical threshold denoising procedure,it employs new threshold function to retain or shrink the wavelet coefficients derived from decomposition,and then reconstructs the image using new wavelet coefficient derived.Different types of noise are added to verify the improved algorithm.Experimental results show that the new threshold function denoising algorithm is more effective than ordinary ones,and the image obtained from new threshold function denoising has greater PSNR and smaller MSE.The value of PSNR improved 1.97 times and the one of MSE declined by 54.4%.