harris角點檢測與ncc匹配（Harris point detection and NCC match）

harris角點檢測與 ncc匹配（Harris point detection and NCC match）harris角點檢測與 ncc匹配(Harris point detection and NCC match) Harris corner detection and NCC matching File1:- Function, y1, Y2, R, c=harris (X) % corner detection using the Harris algorithm % output is an image % result, CNT, R, c=harris (X) % CLC, clear all; % filename=qiao1.bmp; % X= imread (filename.bmp);% reads the image % Info=imfinfo (filename);%; this is customary % f=rgb2gray (X); F=X; % Ori_im=double (f) /255;%unit8 is converted to 64 double precision double64 FX = -2, -1 01, 2,% X direction gradient operator (for Harris corner extraction algorithm) Ix = Filter2 (FX, ori_im);% x directional filtering is good at using filter % FY = 5 85; 000; -5, -8, -5;% Gauss function, first order differential, Y direction (for improved Harris corner extraction algorithm) FY = -2; -1; 0; 1; 2;% Y direction gradient operator (used in Harris corner extraction algorithm) Iy = Filter2 (FY, ori_im);% y directional filtering Ix2 = Ix.2; Iy2 = Iy.2; Ixy = Ix.*Iy; Clear Ix; Clear Iy =% elimination variable ha H=, fspecial (gaussian, 10, 10, 2);% produces Gauss window functions of 7*7, sigma=2 Ix2 = Filter2 (h, Ix2); Iy2 = Filter2 (h, Iy2); Ixy = Filter2 (h, Ixy);% Gauss filtering, respectively Height = size (ori_im, 1); Width = size (ori_im, 2); Result = zeros (height, width);% record corner position, corner value is 1, background is black ha R = zeros (height, width); Rmax = 0;% the maximum R value in the image in order to set the threshold For I = 1:height For J = 1:width M = Ix2 (I, J), Ixy (I, J); Ixy (I, J), Iy2 (I, J);%2*2 matrix R (I, J) = det (M) -0.06* (trace (M) 2;% calculates R and obtains RMAX, which appears to be the whole, and the corner response function If R (I, J) Rmax Rmax = R (I, J); End; End; End; CNT = 0;% record count For I = 2:height-1 For J = 2:width-1% carry on non maximum suppression, window 3*3 If R (I, J) 0.01*Rmax CNT = cnt+1; End; End; End; % % i=1; % for j=1:height % for k=1:width % if result (J, K) =1; % corners1 (I, 1) =j; % corners1 (I, 2) =k; % i=i+1; % end; % end; % end; posr, posc = find (result = 1); % corner number % imshow (ori_im)% and X have the same effect % hold on; % plot (posr, POSC,r.); Y1=result; Y2=cnt; R=posr; c=posc; Return; File2- Function, res=match (A1, cnt1, R1, C1, A2, CNT2, R2, C2) % res=match (A1, A2) % will find the best match point in A2 from A1 and get the res extracted from A2, that is, one-way search % result1, cnt1, R11, c11=harris (A1); CNT2, R22, result2%, c22=harris (A2);% can ensure to match what is, what % figure; % imshow (result1); title (result1 corner position); % figure; title (result2 corner position); % imshow (result2); Win=1/9 1/9 1/9; 1/9 1/9 1/9; 1/9 1/9 1/9; U1=filter2 (win, A1); U2=filter2 (win, A2);% for mean Research on image registration algorithm based on point feature A1=double (A1); A2=double (A2); A=filter2 (win, (a1-u1).2)% for variance B=filter2 (win, (a2-u2).2); m1, n1=size (A1); m2, n2=size (A2); Res1=zeros (M1, N1); Res2=zeros (m2, N2);% searched for matching points For s=1:cnt1 Max=0; p=0; q=0; i=r1 (s, 1); j=c1 (s, 1);%p.q stores coordinates For v=1:cnt2 M=r2 (V, 1); n=c2 (V, 1); K1= (A1 (i-1, J-1) -u1 (I, J) * (A2) (m-1, n-1) -u2 (m, n);% with result is to find what information, 3*3 for such a sparse point of no use K2= (A1 (i-1, J), -u1 (I, J) * (A2 (m-1, n) -u2 (m, n); K3= (A1 (i-1, j+1), -u1 (I, J) * (A2 (m-1, n+1) -u2 (m, n)% is better with loop K4= (A1 (I, J-1), -u1 (I, J) * (A2 (m, n-1) -u2 (m, n); K5= (A1 (I, J), -u1 (I, J) * (A2 (m, n) -u2 (m, n); K6= (A1 (I, j+1), -u1 (I, J) * (A2 (m, n+1) -u2 (m, n); K7= (A1 (i+1, J-1), -u1 (I, J) * (A2 (m+1, n-1) -u2 (m, n); K8= (A1 (i+1, J), -u1 (I, J) * (A2 (m+1, n) -u2 (m, n); K9= (A1 (i+1, j+1), -u1 (I, J) * (A2 (m+1, n+1) -u2 (m, n); Num=k1+k2+k3+k4+k5+k6+k7+k8+k9; Den=sqrt (A (I, J) *B (m, n); Ncc=num/den; If nccmax Max=ncc; p=m; q=n; End End Res2 (P, q) =1; End % can not do it, you can search for a point first Res=res2; Return File3- % The matching of% characteristic points is mainly based on the detection of Harris corner points and match one-way matching function % is suitable for images with white edges, because, when windowed filtering, there is no limit to the extent of Kazakhstan, as far as possible to ensure that the corner is not on the edge CLC, clear, all; A1=imread (qiao1.bmp); A2=imread (qiao2.bmp); The role of%double is great Subplot (1,2,1); imshow (A1); subplot (1,2,2); imshow (A2); title (original image); result1, cnt1, R1, c1=harris (A1);% corner detection, the raw focus position map result is obtained result2, CNT2, R2, c2=harris (A2); Figure; subplot (1,2,1); imshow (A1); hold, on; plot (C1, R1,g.); Subplot (1,2,2); imshow (A2); hold, on; plot (C2, R2,g.); title ( graph 1, 2 corner graph); Res2=match (A1, cnt1, R1, C1, A2, CNT2, R2, C2);% starts from result1 and searches possible in result2 r22, c22=find (res2=1); m22, n22=size (R22); Cnt22=m22; Res1=match (A2, cnt22, R22, C22, A1, cnt1, R1, C1);% reverse search res2-result1 Res1=and (res1, result1);% guarantees that the reverse matching does not appear to be impossible r11, c11=find (res1=1); m11, n11=size (R11); Cnt11=m11; Res22=match (A1, cnt11, R11, C11, A2, cnt22, R22, C22);% starts from res1 and searches possible in res2 Res22=and (res22, res2); r222, c222=find (res22=1); m222, n222=size (r222); Cnt222=m222; Of course, the more the number of%, the number will be less and less, the match was always relatively Kazakhstan The end result of the% attention points are not necessarily the same as the ha, there is no need to