资源简介
piotr_toolbox,作者网址打不开了,包括视频和图像的处理,关键是seq视频文件找了很多工具都无法打开,该工具可以打开。
代码片段和文件信息
function chns = chnsCompute( I varargin )
% Compute channel features at a single scale given an input image.
%
% Compute the channel features as described in:
% P. Doll醨 Z. Tu P. Perona and S. Belongie
% “Integral Channel Features“ BMVC 2009.
% Channel features have proven very effective in sliding window object
% detection both in terms of *accuracy* and *speed*. Numerous feature
% types including histogram of gradients (hog) can be converted into
% channel features and overall channels are general and powerful.
%
% Given an input image I a corresponding channel is a registered map of I
% where the output pixels are computed from corresponding patches of input
% pixels (thus preserving overall image layout). A trivial channel is
% simply the input grayscale image likewise for a color image each color
% channel can serve as a channel. Other channels can be computed using
% linear or non-linear transformations of I various choices implemented
% here are described below. The only constraint is that channels must be
% translationally invariant (i.e. translating the input image or the
% resulting channels gives the same result). This allows for fast object
% detection as the channels can be computed once on the entire image
% rather than separately for each overlapping detection window.
%
% Currently three channel types are available by default (to date these
% have proven the most effective for sliding window object detection):
% (1) color channels (computed using rgbConvert.m)
% (2) gradient magnitude (computed using gradientMag.m)
% (3) quantized gradient channels (computed using gradientHist.m)
% For more information about each channel type including the exact input
% parameters and their meanings see the respective m-files which perform
% the actual computatons (chnsCompute is essentially a wrapper function).
% The converted color channels serve as input to gradientMag/gradientHist.
%
% Additionally custom channels can be specified via an optional struct
% array “pCustom“ which may have 0 or more custom channel definitions. Each
% custom channel is generated via a call to “chns=feval(hFuncIpFunc{:})“.
% The color space of I is determined by pColor.colorSpace use the setting
% colorSpace=‘orig‘ if the input image is not an ‘rgb‘ image and should be
% left unchanged (e.g. if I has multiple channels). The input I will have
% type single and the output of hFunc should also have type single.
%
% “shrink“ (which should be an integer) determines the amount to subsample
% the computed channels (in applications such as detection subsamping does
% not affect performance). The params for each channel type are described
% in detail in the respective function. In addition each channel type has
% a param “enabled“ that determines if the channel is computed. If
% chnsCompute() is called with no inputs the output is the complete
% default params (pChns). Otherwise the outputs are the comp
属性 大小 日期 时间 名称
----------- --------- ---------- ----- ----
目录 0 2013-10-15 09:48 toolbox\
目录 0 2013-10-15 09:48 toolbox\channels\
文件 9488 2013-09-29 06:54 toolbox\channels\chnsCompute.m
文件 10780 2013-09-29 06:54 toolbox\channels\chnsPyramid.m
文件 5201 2013-09-29 06:54 toolbox\channels\chnsScaling.m
文件 1765 2013-09-29 06:54 toolbox\channels\Contents.m
文件 3204 2013-09-29 06:54 toolbox\channels\convBox.m
文件 2441 2013-09-29 06:54 toolbox\channels\convMax.m
文件 4560 2013-09-29 06:54 toolbox\channels\convTri.m
文件 3098 2013-09-29 06:54 toolbox\channels\fhog.m
文件 1310 2013-09-29 06:54 toolbox\channels\gradient2.m
文件 3558 2013-09-29 06:54 toolbox\channels\gradientHist.m
文件 2455 2013-09-29 06:54 toolbox\channels\gradientMag.m
文件 3597 2013-09-29 06:54 toolbox\channels\hog.m
文件 1360 2013-09-29 06:54 toolbox\channels\hogDraw.m
文件 2349 2013-09-29 06:54 toolbox\channels\imPad.m
文件 2446 2013-10-15 09:29 toolbox\channels\imResample.m
目录 0 2013-10-15 09:48 toolbox\channels\private\
文件 2655 2013-09-29 06:54 toolbox\channels\private\chnsTestCpp.cpp
文件 10613 2013-10-15 09:30 toolbox\channels\private\convConst.cpp
文件 18665 2013-09-29 06:54 toolbox\channels\private\convConst.mexa64
文件 29184 2013-09-19 13:25 toolbox\channels\private\convConst.mexw64
文件 19354 2013-10-15 09:30 toolbox\channels\private\gradientMex.cpp
文件 23054 2013-09-29 06:54 toolbox\channels\private\gradientMex.mexa64
文件 38912 2013-09-29 06:51 toolbox\channels\private\gradientMex.mexw64
文件 5609 2013-09-29 06:54 toolbox\channels\private\imPadMex.cpp
文件 28577 2013-09-29 06:54 toolbox\channels\private\imPadMex.mexa64
文件 40960 2013-09-29 06:54 toolbox\channels\private\imPadMex.mexw64
文件 8042 2013-09-29 06:54 toolbox\channels\private\imResampleMex.cpp
文件 21350 2013-09-29 06:54 toolbox\channels\private\imResampleMex.mexa64
文件 34304 2013-09-05 07:56 toolbox\channels\private\imResampleMex.mexw64
............此处省略648个文件信息
评论
共有 条评论