CAJ | 학술논문

由于图像软组织的对比度和信噪比均较低的原因，传统的二进小波算法在进行磁共振图像的增强处理时效果不明显。文章首先在分析小波变换对噪声影响规律的基础上，结合磁共振图像( MRI)特点，对二进小波变换中各尺度上小波系数间的相关性进行分析和归类。其次，文章通过改进的小波系数相关性置信度方法，将MRI图像的像素点分为信号点、噪声点和性质未定点，并采用不同的增强函数对信号点和噪声点进行有效处理。最后，文章提出了一个基于三阶分段函数的性质未定点处理算法。实验表明，文章中所采取的方法能较好地实现像素点的分类，抑制其中的噪声点，有效地增强目标像素点。
유우도상연조직적대비도화신조비균교저적원인，전통적이진소파산법재진행자공진도상적증강처리시효과불명현。문장수선재분석소파변환대조성영향규률적기출상，결합자공진도상( MRI)특점，대이진소파변환중각척도상소파계수간적상관성진행분석화귀류。기차，문장통과개진적소파계수상관성치신도방법，장MRI도상적상소점분위신호점、조성점화성질미정점，병채용불동적증강함수대신호점화조성점진행유효처리。최후，문장제출료일개기우삼계분단함수적성질미정점처리산법。실험표명，문장중소채취적방법능교호지실현상소점적분류，억제기중적조성점，유효지증강목표상소점。
Dyadic wavelet transforms have no demonstrable effect on MRI enhancement,because MRI has the disadvantage of its lower contrast and noise-signal ratio. After analyzing the effects of wavelet on signal noise and based on the characteristics of MRI,this paper analyzes the correlation of dyadic wavelet transform of wavelet coefficient in each scale and classes them to different types. Then a revised method of wavelet coefficient confidence is introduced,hence MRI pixels are classified to three types, including singal itself, noise and undefined pot. At last, after using different functions on enhance steps for the first two types of signals,a new piecewise function has been raised to deal with the undefined pot. Experiment shows that using this method can effectively classify pixels,reduce the noise and enhance the true signal pixel.