renlianshibie-face.rar_face video matlab_recognize video_人脸识别 视频

% By lyqmath % DLUT School of Mathematical Sciences % BLOG：http://blog.csdn.net/lyqmath clc; clear all; close all; % 载入图像 Img = imread('face.gif'); if ndims(Img) == 3 I=rgb2gray(Img); else I = Img; end BW = im2bw(I, graythresh(I)); % 二值化 figure; subplot(2, 2, 1); imshow(Img); title('原图像', 'FontWeight', 'Bold'); subplot(2, 2, 2); imshow(Img); title('网格标记图像', 'FontWeight', 'Bold'); hold on; [xt, yt] = meshgrid(round(linspace(1, size(I, 1), 10)), ... round(linspace(1, size(I, 2), 10))); mesh(yt, xt, zeros(size(xt)), 'FaceColor', ... 'None', 'LineWidth', 3, ... 'EdgeColor', 'r'); subplot(2, 2, 3); imshow(BW); title('二值图像', 'FontWeight', 'Bold'); [n1, n2] = size(BW); r = floor(n1/10); % 分成10块，行 c = floor(n2/10); % 分成10块，列 x1 = 1; x2 = r; % 对应行初始化 s = r*c; % 块面积 for i = 1:10 y1 =1; y2 = c; % 对应列初始化 for j = 1:10 if (y2<=c || y2>=9*c) || (x1==1 || x2==r*10) % 如果是在四周区域 loc = find(BW(x1:x2, y1:y2)==0); [p, q] = size(loc); pr = p/s*100; % 黑色像素所占的比例数 if pr <= 100 BW(x1:x2, y1:y2) = 0; end end y1 = y1+c; % 列跳跃 y2 = y2+c; % 列跳跃 end x1 = x1+r; % 行跳跃 x2 = x2+r; % 行跳跃 end [L, num] = bwlabel(BW, 8); % 区域标记 stats = regionprops(L, 'BoundingBox'); % 得到包围矩形框 Bd = cat(1, stats.BoundingBox); [s1, s2] = size(Bd); mx = 0; for k = 1:s1 p = Bd(k, 3)*Bd(k, 4); % 宽*高 if p>mx && (Bd(k, 3)/Bd(k, 4))<1.8 % 如果满足面积块大，而且宽/高<1.8 mx = p; j = k; end end subplot(2, 2, 4);imshow(I); hold on; rectangle('Position', Bd(j, :), ... 'EdgeColor', 'r', 'LineWidth', 3); title('标记图像', 'FontWeight', 'Bold');