Face / Object Detection－天天向上

Viola–Jones object detection[1][7]

包含下列四個步驟

1. Haar Feature Selection
2. Creating an Integral Image
3. Adaboost Training
4. Cascading Classifiers

MATLAB File Exchange有Viola-Jones的改良版,

主要增加Haar Like Feature的核心(kernel)選擇不同

請先下載

>> mex -setup

Welcome to mex -setup.  This utility will help you set up  a default compiler.  For a list of supported compilers, see  http://www.mathworks.com/support/compilers/R2012b/win64.html   Please choose your compiler for building MEX-files:   Would you like mex to locate installed compilers [y]/n? y  Select a compiler: [1] Microsoft Visual C++ 2010 in C:\Program Files (x86)\Microsoft Visual Studio 10.0   [0] None   Compiler: 1  Please verify your choices:   Compiler: Microsoft Visual C++ 2010  Location: C:\Program Files (x86)\Microsoft Visual Studio 10.0    Are these correct [y]/n? y

--------------------------------------------------------------------

>> setup_fdt

Unzipping negatives picts ... Compile mex files ... compiling chlbp compiling chlbp_adaboost_binary_train_cascadecompiling chlbp_adaboost_binary_predict_cascade compiling chlbp_gentleboost_binary_train_cascade compiling chlbp_gentleboost_binary_predict_cascade compiling detector_haar compiling detector_mblbp compiling eval_chlbp compiling eval_haar compiling eval_haar_subwindow compiling eval_mblbp compiling eval_mblbp_subwindows compiling haar compiling haar_ada_weaklearner compiling haar_adaboost_binary_train_cascade compiling haar_adaboost_binary_predict_cascade compiling haar_featlist compiling haar_gentle_weaklearner compiling haar_gentleboost_binary_train_cascade compiling haar_gentleboost_binary_predict_cascade compiling haar_scale compiling imresize compiling area compiling homkermap compiling homkertable compiling eval_hmblbp_spyr_subwindow compiling detector_mlhmslbp_spyr compiling eval_hmblgp_spyr_subwindowcompiling detector_mlhmslgp_spyr compiling mblbp compiling mblbp_ada_weaklearner compiling mblbp_adaboost_binary_train_cascade compiling mblbp_adaboost_binary_predict_cascade compiling mblbp_featlist compiling mblbp_gentle_weaklearner compiling mblbp_gentleboost_binary_train_cascade compiling mblbp_gentleboost_binary_predict_cascade compiling rgb2gray compiling fast_rotate compiling haar_ada_weaklearner_memory compiling haar_adaboost_binary_predict_cascade_memory compiling haar_adaboost_binary_train_cascade_memory compiling haar_gentle_weaklearner_memory compiling haar_gentleboost_binary_predict_cascade_memory compiling haar_gentleboost_binary_train_cascade_memory compiling int8tosparse compiling fast_haar_ada_weaklearner compiling fast_haar_adaboost_binary_train_cascade compiling train_dense Failed to compile mex-files, unzip precompiled mex64 Error using checkfilename>validateFilename (line 166) Function UNZIP was unable to find file ''mexw64.zip''. Error in checkfilename (line 50) [fullfilename, fid] = validateFilename( ... Error in parseUnArchiveInputs (line 82) [archiveFilename, url] = checkfilename(archiveFilename, validExtensions, ... Error in unzip (line 58) [zipFilename, outputDir, url, urlFilename] = parseUnArchiveInputs( ... Error in setup_fdt (line 80)             unzip('mexw64.zip' , conf.path);

--------------------------------------------------------------------

demo_haar.m

clear, close all,clc,drawnowload viola_24x24

載入後觀察X, y毫無頭緒,

根據兩者維度推測, 嘗試 imshow(X(:,:,1))

原來裡面塞了一堆faces

imshow(X(:,:,2))

>> plot(y)

各位看了這個下面這張圖想必猜得出來1~4915筆前的資料是一群,

4916~12788筆的資料是另一群

>> imshow(X(:,:,4917))

真的是non-face, BINGO!!!

特徵擷取

A: Dims=1x2, kernelA=[-1, 1];

B: Dims=2x1, kernelB=[1; –1];

C: Dims=1x3, kernelC=[-1, 1, –1];

D: Dims=2x2, kernelD=[-1, 1; 1 –1];

private static int CalCounts(int frameSize)  {      const int features = 5;      // All five feature types:      // { width, height} for each set      int[,] feature = new int[features, 2]{ { 2, 1 }, { 1, 2 }, { 3, 1 }, { 1, 3 }, { 2, 2 } };         int count = 0;      int i, sizeX, sizeY, x, y, width, height;      // Each feature:      for (i = 0; i < features; i++) {          sizeX = feature[i,0];          sizeY = feature[i,1];                // Each position:          for (x = 0; x <= frameSize-sizeX; x++) {              for (y = 0; y <= frameSize-sizeY; y++) {              // Each size fitting within the frameSize:                  for (width = sizeX; width <= frameSize-x; width+=sizeX) {                      for (height = sizeY; height <= frameSize-y; height+=sizeY) {                      count++;                      }                  }              }          }      }      return count;  }