《基于最小二乘影像匹配的畸变图像矫正算法研究王岩松》由会员分享，可在线阅读，更多相关《基于最小二乘影像匹配的畸变图像矫正算法研究王岩松（4页珍藏版）》请在金锄头文库上搜索。
1、cI|:1000-1506(2002)01-0015-04Kl=MmE王岩松,阮秋琦(ZYv9/,100044)K1:针对高精度的畸变图像,提出了一种基于最小二乘影像匹配的高精度畸变图像矫正算法.算法首先利用特征提取与边缘检测对图像进行预处理,并且将特征匹配与最小二乘算法相结合,从而实现了图像与模板之间精确的子像素定位与匹配.实验表明,该算法较好的解决了目前高精度畸变图像矫正算法中普遍存在的定位和匹配精度较差的缺陷,图像矫正效果良好,是一种有效的畸变图像矫正算法.1oM:影像匹配;最小二乘算法;边缘检测;模板图像ms|:TP391.41DSM:AA Distorted Image Rectif
2、ication AlgorithmBased on Least Square Image MatchingWANG Yan-song , RUAN Qiu-qi(College of Computer and Information Technology, Northern Jiaotong University, Beijing 100044,China)Abstract:This paper presents a distorted image rectification algorithm which is based on leastsquare image matching.This
3、 algorithm first preprocesses the image with the method of character-extraction and edge-detection, then links the character matching with least square algorithm toachieve the exact sub-pixel locating and matching between original image and templet image.Ex-periments show that this algorithm can gre
4、atly improve the precision of image locating and match-ing, and it is a better algorithm in distorted image rectification.Key words:image matching;least square algorithm;edge detection;templet imagemMm)L=TR51 .-1Em、n、+Z.vsE(H6+,H(?sm+,y7rTD,H1l1 .SBZE,5VV9j$3?%EJ.ZE+,7O1WM1.ZE135f/,M1+1.Mm,4Kl=mE.E1
5、ay$m+4|H_,iO|+Kl=EMMm.LEV,Ez%ERiJ,mH+61z,7Oy、F.2?+.l:2001-09-30Te:(1977),3,V3.email:2612002M2ZYvJOURNAL OF NORTHERN JIAOTONG UNIVERSITY Vol.26 No.1Feb.20021V3MYM1.9j$3BZE.7Kl=MmEXVm?,1ayV,n5m0m+4|H_,Kl=,V7LCmWMm.E1+4|、H_、Kl=Z.+4|H_4|m0mZ_m,4|0+,_mH,|mL=mKl=,mMsMvl,rmM.Mm,MmEmnm1.m1Em1.1+4|H_+4|4|m0m
6、Z_m,LH_m.Mm,+H0Z_s0vTH.ym+HiBmv),7Bm“M”Z,90Kv).yN4|0m+4|1.mf(x,y)(x,y)0_lg(f(x,y)= f x, f yT(1)T|mf(x,y)=c1 x +c2y +c3 (2)T,c1、c2、c3M.|m,5/T= (x,y)s f(x,y)-c1x -c2y -c32 min (3)Vpc1c2,iOVp0vlg = c21 +c22,=arctan c2c1(4)T(4)Vm0mZ_m,Haralick;4BT2mE,i0Z_pvTH_5.1.2ymMyZ4m$.mHVf,i8InT.Bd9,3LVd9(E(E3 .1.3
7、Kl=dKl=s#E.Kl=|Kl=E+3M,V7PL=r0.1.+4|e,BVr1P.13L|,B1pr1/4.7Kl=(LSM)BrE,YVVPL=r0.1.|+Kl=E,Kl=.16ZYv26m2Kl=Vm+H,L=mROI(Region of Interest,u)mvM,In+Mf/,|+MA_M,4G1(x,y)+n1(x,y)=S G2(a0 +a1 x +a2y,b0 +b1 x +b2y)+n2(x,y) (5),G1L=m+ROI,G2ROI,SL,n1n2m.2,ai、bi(i=0,1,2)sY_M.|TL,VZV =c1dS +c2da0 +c3da1 +c4da2 +c5db0 +c6db1 +c7db2 -(G1 -G2) (6)ci(i=1,7)M.ROIu=yZ,7s9,VTyEZT/CT CX =CTG (7)NZVp+M.Kl=+$BpV,EdLCmnm2.2LTsL400200
