facedetection
所属分类:matlab编程
开发工具:matlab
文件大小:700KB
下载次数:2
上传日期:2012-12-18 20:56:26
上 传 者:
wotaiyangde
说明: A novel and effient face detection for some reseacher.
文件列表:
facedetection (0, 2007-11-06)
facedetection\aggregatePoses.m (2063, 2006-06-04)
facedetection\checkEdgeLocation.m (1860, 2006-05-12)
facedetection\checkEdgeScale.m (1681, 2006-06-04)
facedetection\createFilter.m (1188, 2006-05-09)
facedetection\demo.m (3169, 2006-06-04)
facedetection\detectFaces.m (2512, 2006-06-03)
facedetection\drawFace.m (1080, 2006-06-04)
facedetection\drawFaces.m (538, 2006-06-04)
facedetection\drawRectangle.m (600, 2006-05-09)
facedetection\firstRun-2.mat (209315, 2006-05-17)
facedetection\firstRun.m (2033, 2006-05-16)
facedetection\fullRun-2.mat (452811, 2006-05-24)
facedetection\fullRun.m (2168, 2006-06-01)
facedetection\getEdgeMap.m (4381, 2006-05-17)
facedetection\getFlatMap.m (1191, 2006-05-14)
facedetection\getPoseImage.m (1972, 2006-05-14)
facedetection\getPoseLevels.m (746, 2006-05-17)
facedetection\getPoseRelatives.m (618, 2006-05-17)
facedetection\getPoses.m (4861, 2006-05-17)
facedetection\getPoseTrainingSet.m (2252, 2006-05-16)
facedetection\integralFilter.m (1177, 2006-05-09)
facedetection\integralImage.m (266, 2006-05-09)
facedetection\integralPoint.m (518, 2006-05-09)
facedetection\labelImages.m (2328, 2006-05-16)
facedetection\normalizeImages.m (1308, 2006-05-13)
facedetection\sampleImage.jpg (25363, 2006-05-24)
facedetection\test.m (7713, 2006-06-03)
facedetection\testCell.m (1776, 2006-05-17)
facedetection\testImage.m (4732, 2006-06-03)
facedetection\testWindow.m (2732, 2006-06-03)
facedetection\trainCell.m (4881, 2006-05-16)
facedetection\trainCells.m (2663, 2006-05-17)
Fast Face Detection
-------------------
Authors:
-Mohamed ALY
-Jonathan LEE
Contents
---------
-*.m source files
-.mat trained classifiers
-sample image
-example file
Source Files
------------
-aggregatePoses: aggregates all the face detections returned by
detectFaces.m in an attempt to eliminate false alarms
-checkEdgeLocation: checks if a specific edge (location, type, scale) is
present in the input edge map
-checkEdgeScale: checks the whole edge map for specific edge type at
specific scale (all locations)
-detectFaces: checks a given image for faces at different subsampled scales
of the image (calls testImage for different scales)
-drawFace: draws the input faces detections on the current figure (used to
show the triangles of the detected faces)
-firstRun: used to train the classifier for the 32x32 window size
-fullRun: used to train the classifier for the ***x*** window size
-getEdgeMap: generates an edge map for the input image
-getFlatMap: generates a flat area map fr the input image
-getPoseImage: generates an image with a given pose from an input image
(used in the training process to generate the synthetic training set)
-getPoseLevels: gets the depth of the pose tree
-getPoseRelations: used to get the children and parents of a given pose in
the pose tree (used in detection process to navigate the tree)
-getPoses: used to generate the pose tree according to the number of splits
required and their types (splits on location, scale or tilt)
-getPoseTrainingSet: used to generate the synthetic training set for each
pose from the original input training set by generating different
random orientation, scales and translations
-labelImages: used in the normalization of the original input traininig set
(ORL database) so label the locations of the eyes and mouth center
-normalizeImages: uses the labeled images to normalize the input training
set so that all faces have eyes and mouth at the same position (to make
generating the synthetic training set systematic)
-testCell: used to test an input window for faces according to a specific
cell in the pose tree i.e. tests for the number of edges and flat areas
in that window and compare to the threshold got from training
-testImage: used detect a given image for faces at the same scale of the
input image
-testWindow: tests a given window of the image going down the pose tree
in a breadth first approach, such that it doesnt test a given cell
unless its parent responds true
-trainCell: the main training algorithm that trains the cell classifier of
a given cell in the pose tree
-trainCells: called by firstRun or fullRun to generate a subset of cell
classifiers
Mat files
----------
-firstRun-2.mat: classifier trained using 32x32 window & 8x8 blocks with a
pose tree of 5 levels = 245 classifiers (2 splits on location, 1 on
scale, 1 on tilt, and 1 on scale)
-fullRun-2.mat: classifier trained using ***x*** window & 8x8 blocks with a
pose tree of 6 levels = 501 classifiers (2 splits on location, 1 on
scale, 1 on tilt, 1 on scale, and 1 on tilt again)
-trainSet.mat: the normalized training set of 400 images from ORL database
Sample Image
------------
sampleImage.jpg
Example script
--------------
demo.m
近期下载者:
相关文件:
收藏者: