CAJ | 학술논문

递推最小二乘 RLS（Recursive of Least Square）算法是自适应滤波算法中的精确分析算法。它具有收敛速率快，精确度高等特点，但是发现目前 RLS 算法多用于对一维信号的去噪处理。使用递推最小二乘（RLS）算法对二维图像进行去噪，从处理一维信号变成处理二维图像信号，需要对 RLS 算法进行改进。先迭代得到滤波器参数，形成3×3滤波掩模，再改进算法对图像进行滤波；同时与常数比率维纳滤波和自相关函数的维纳滤波算法的去噪效果进行对比。结论证明在对图像进行较严重的模糊和加噪处理后，其他两种算法对图像的还原能力差，而递推最小二乘自适应滤波（RLS）算法具有优良的图像去噪性能。
체추최소이승 RLS（Recursive of Least Square）산법시자괄응려파산법중적정학분석산법。타구유수렴속솔쾌，정학도고등특점，단시발현목전 RLS 산법다용우대일유신호적거조처리。사용체추최소이승（RLS）산법대이유도상진행거조，종처리일유신호변성처리이유도상신호，수요대 RLS 산법진행개진。선질대득도려파기삼수，형성3×3려파엄모，재개진산법대도상진행려파；동시여상수비솔유납려파화자상관함수적유납려파산법적거조효과진행대비。결론증명재대도상진행교엄중적모호화가조처리후，기타량충산법대도상적환원능력차，이체추최소이승자괄응려파（RLS）산법구유우량적도상거조성능。
Recursive least squares (RLS)algorithm is an accurate analysis algorithm among adaptive filtering algorithms.It has the characteristics of fast convergence rate and high precision.But we find that current RLS algorithm is more used in removing the noise of one-dimension signals.Using recursive least squares algorithm for 2D image denoising,and converting the processing of one dimensional single to two dimensional image signal,there has the need to improve the RLS algorithm.First,we obtain filter parameters with iteration,and in turn form the 3 ×3 filter mask;then we improve the algorithm to filter the image;At the same time,we compare its denoising effect with that of the constant ratio Wiener filtering and the autocorrelation function Wiener filtering algorithm.The conclusion proves that after applying heavier blurring and noising treatment on the image,other two kinds of algorithms have poorer restoration ability on the image,and the recursive least squares adaptive filtering algorithm has good image denoising performance.