HoG 特征可视化
来源:互联网 发布:淘宝店铺在线布局 编辑:程序博客网 时间:2024/06/06 01:12
function [Feats,Defaults,Values]= features(f);% USERLOCAL/FEATURES% Copyright 2000-2005 The MathWorks, Inc. and Ford Global Technologies, Inc.% $Revision: 1.2.2.4 $ $Date: 2011/10/01 20:58:22 $lab= labels(f);np = length(lab);lab= lab(:)';func=lab;dgfunc= lab;% parameters as response featuresfor i=1:length(lab); lab{i}= strrep(lab{i},'\',''); func{i}= sprintf('p(%d)',i); dgfunc{i}= sprintf('delparam(f,%d)',i);endFeats= struct('Display',lab,... 'Function',func,... 'delG',dgfunc,... 'Name',lab,... 'IsDatum',0,... 'index',num2cell(1:np),...'IsLinear',1);try % possible userdefined [rf,dGl]= rfvals(f.userdefined); [rnames,Defaults]= rfnames(f.userdefined); if ~isempty(rnames) && (~iscellstr(rnames) || length(rf)~=length(rnames)) error(message('mbc:xregtransient:InvalidFile')) end for i= 1:length(rf) Feats(np+i)= struct('Display',rnames{i},... 'Function',sprintf('rfvals(f,%d)',i),... 'delG',sprintf('delrf(f,%d)',i),... 'Name',rnames{i},... 'IsDatum',0,... 'index',np+i,... 'IsLinear',0); endcatch error(message('mbc:xregtransient:InvalidFile'))endif nargout==3 Values= zeros(np,nfactors(f));end
显示函数:
% showHOG(w)%% Legacy HOG visualizationfunction out = showHOG(w)w = w(:, :, 1:9) + w(:, :, 10:18) + w(:, :, 19:27);w = w / 3;w = repmat(w, [1 1 3]);w = padarray(w, [0 0 5], 'post');% Make pictures of positive and negative weightsbs = 20;pos = HOGpicture(w, bs);neg = HOGpicture(-w, bs);% Put pictures together and drawbuff = 10;if min(w(:)) < 0 pos = padarray(pos, [buff buff], 0.5, 'both'); neg = padarray(neg, [buff buff], 0.5, 'both'); im = [pos neg];else im = pos;endim(im < 0) = 0;im(im > 1) = 1;if nargout == 0, imagesc(im); colormap gray; axis image;else, out = im;end% HOGpicture(w)%% Make picture of positive HOG weights.% im = HOGpicture(w, bs)function im = HOGpicture(w, bs)if ~exist('bs', 'var'), bs = 20;end% construct a "glyph" for each orientaionbim1 = zeros(bs, bs);bim1(:,round(bs/2):round(bs/2)+1) = 1;bim = zeros([size(bim1) 9]);bim(:,:,1) = bim1;for i = 2:9, bim(:,:,i) = imrotate(bim1, -(i-1)*20, 'crop');end% make pictures of positive weights bs adding up weighted glyphss = size(w); w(w < 0) = 0; im = zeros(bs*s(1), bs*s(2));for i = 1:s(1), iis = (i-1)*bs+1:i*bs; for j = 1:s(2), jjs = (j-1)*bs+1:j*bs; for k = 1:9, im(iis,jjs) = im(iis,jjs) + bim(:,:,k) * w(i,j,k+18); end endendscale = max(max(w(:)),max(-w(:)));im = im / scale;
效果图:
1 1
- HOG特征显示 可视化HOG
- HoG 特征可视化
- HOG特征的可视化
- HOG特征
- HOG特征
- HOG特征
- HOG特征
- hog 特征
- HOG特征
- HOG特征
- HOG特征
- HOG特征
- HOG特征
- HOG特征
- HOG特征
- HOG特征
- hog 特征
- HOG特征
- Eclipse下编写C++程序——CDT环境搭建
- Linux AIO基础
- dt.jar和tools.jar calsspath
- 1501102328-蓝桥杯-算法训练 2的次幂表示
- web移动端滑动
- HoG 特征可视化
- OC反射机制一
- 软件设计模式---创建型模式(一)
- 【最小割】【bzoj 3511】: 土地划分
- 【BZOJ】【P2743】【HEOI2012】【采花】【离线+树状数组】
- 黄聪:Python 字符串操作(string替换、删除、截取、复制、连接、比较、查找、包含、大小写转换、分割等)
- Please ensure that adb is correctly located 问题的解决
- Java线程专栏文章汇总
- IOS开发之纯代码界面--基本控件使用篇