adding HOG Features in a matrix "matlab" - matlab

After extracting HOG features of folder of images, I want to add all this results in one matrix. How I can do this? this is my code in matlab:
% Get list of all jpg files in this directory
% DIR returns as a structure array. You will need to use () and . to get
% the file names.
%to get all images in Motorbikes directory
imagefiles = dir('Motorbikes/*.jpg');
nfiles = length(imagefiles); % Number of files found
for ii=1:nfiles
currentfilename = ['Motorbikes/',imagefiles(ii).name];
currentimage = imread(currentfilename);
images{ii} = currentimage;
end
%To extract features of each image
featureMatrix = [];
for jj=1:nfiles
currentImageFeatures = extractHOGFeatures(images{jj});
featureMatrix = [featureMatrix, currentImageFeatures];
end

Related

Recursively read images from subdirectories

I am stuck on something that is supposed to be so simple.
I have a folder, say main_folder with four sub folders, say sub1, sub2, sub3 and sub4 each containing over 100 images. Now am trying to read and store them in an array. I have looked all over the internet and some MATLAB docs:
here, here and even the official doc.
My code is like this:
folder = 'main_folder/**'; %path containing all the training images
dirImage = dir('main_folder/**/*.jpg');%rdir(fullfile(folder,'*.jpg')); %reading the contents of directory
numData = size(dirImage,1); %no. of samples
arrayImage = zeros(numData, 133183); % zeros matrix for storing the extracted features from images
for i=1:numData
ifile = dirImage(i).name;
% ifolder = dirImage(i).folder;
I=imread([folder, '/', ifile]); %%%% read the image %%%%%
I=imresize(I,[128 128]);
...
If I try the code in the above snippet, the images are not read.
But if I replace the first two lines with something like:
folder = 'main_folder/'; %path containing all the training images
dirImage = dir('main_folder/sub1/*.jpg'); %rdir(fullfile(folder,'*.jpg'));
then all images in sub1 are read. How can I fix this? Any help will be highly appreciated. I want to read all the images in the four sub folders at once.
I am using MATLAB R2015a.
I believe you will need to use genpath to get all sub-folders, and then loop through each of them, like:
dirs = genpath('main_folder/'); % all folders recursively
dirs = regexp(dirs, pathsep, 'split'); % split into cellstr
for i = 1:numel(dirs)
dirImage = dir([dirs{i} '/*.jpg']); % jpg in one sub-folder
for j = 1:numel(dirImage)
img = imread([dirs{i} '/' dirImage(j).name]);
% process img using your code
end
end

How to edit multiple .mat files in a folder using MATLAB?

I am trying to edit 100 .mat files in a folder with no pattern in their names. Each file contains a matrix of size 100-by-10, and I want to convert them to 10-by-10-by-10. How do I go about this?
Provided you have a single matrix in each file, and want to overwrite the original matrix:
listing = dir('*.mat'); % finds all .mat files in your pwd
for ii = 1:numel(listing)
tmp = listing(ii).name;
s = load(tmp); % load a file
Fname = fieldnames(s); % get matrix name
out = reshape(s.(Fname{1}),[10 10 10]); % reshape
save(tmp,'out') % Save, overwriting original
end

how to insert noise and save multiple images in different folders using a loop?(matlab)

I am new in image processing and I want help. I have a folder (dataset) that contains 1000 images and I want to insert noise 'salt & pepper' with different density noise (0.01,0.02 and 0.03) , I used this line to do this:
im = imread('C:\Users\SAMSUNG\Desktop\AHTD3A0002_Para1.tif');
J = imnoise(im,'salt & pepper',0.01);
Please help me to do this : I want to save the result in 3 folder ( data1 contains images after noise with d=0.01, data2 contains images after noise with d=0.02 and data3 contains images after noise with d=0.03).
any suggestation and thanks in advance
Following code will allow you to select the folder and create the noised pictures in 3 different folders. It will only select the '*.tif' files which you can modify in the code. And if you need to create more noise levels, create a loop to name the folders and files dynamically.
% get dir
folderX = uigetdir();
% get files
picFiles = dir('*.tif');
% loop over the files and save them with the noise
for ii = 1:length(picFiles)
currentIm = imread([folderX, '\', picFiles(ii).name]);
% create folders if not exist
if ~exist([folderX,'\noise_0.01\'], 'dir')
% create folders
mkdir([folderX,'\noise_0.01\']);
end
if ~exist([folderX,'\noise_0.02\'], 'dir')
% create folders
mkdir([folderX,'\noise_0.02\']);
end
if ~exist([folderX,'\noise_0.03\'], 'dir')
% create folders
mkdir([folderX,'\noise_0.03\']);
end
J1 = imnoise(currentIm,'salt & pepper',0.01);
imwrite(J1,fullfile([folderX, '\noise_0.01\', picFiles(ii).name]));
J2 = imnoise(currentIm,'salt & pepper',0.02);
imwrite(J2,fullfile([folderX, '\noise_0.02\', picFiles(ii).name]));
J3 = imnoise(currentIm,'salt & pepper',0.03);
imwrite(J3,fullfile([folderX, '\noise_0.03\', picFiles(ii).name]));
end
An easy solution with 2 for loop.
%save the noise parameter.
noise = [0.01,0.02,0.03];
for i = 1:1000
%we generate the filename (you can adapt this code)
imname = fullfile('C:\Users\SAMSUNG\Desktop\',sprintf('AHTD3A0002_Para%d.tif',i))
%read the image.
im = imread(imname);
for j = 1:length(noise)
%apply the noise
J = imnoise(im,'salt & pepper',noise(j));
%save image in the right folder
imwrite(J,fullfile('C:\Users\SAMSUNG\Desktop',sprintf('folder%d',j)));
end
end

frame to video conversion matlab

So I just started with image processing/computer vision in MATLAB.
So my first task is to convert a series of images(frames) into a video. So I went through online sources (MATLAB website more specifically) to get a way to do it.
So the one that I implemented is http://www.mathworks.com/help/matlab/examples/convert-between-image-sequences-and-video.html which solved the problem for me.
However, when I play it, the video seems jumpy in some places. Like it would bring a different frame in the middle and make the whole video jumpy for that split second. It happens a couple of places in the video.
Any anyone knows why this happens?
Thanks
PS below is the code I use:
myFolder = 'C:\Users\owner\Desktop\MATLAB GUI\Color\Color'; %Specify Directory
filePattern = fullfile(myFolder, '*.jpg') %identify jpg files
jpegFiles = dir(filePattern) %use dir to list jpg files
size = length(jpegFiles); % length of the size of the file
outputVideo = VideoWriter(fullfile(myFolder,'video1.avi'));
outputVideo.FrameRate = 30;
open(outputVideo);
for i = 1:length(jpegFiles) %load all the files in the directory
j = i; %%accumulating the number of jpegfiles into handles.j
baseFileName = jpegFiles(i).name;
fullFileName = fullfile(myFolder, baseFileName);
%fprintf(1, 'Now reading %s\n', fullFileName); %filename of image
imageArray = imread(fullFileName); %image being read
%imageArray = rgb2gray(imageArray);
imagecell{i} = imageArray; %storing the images in imagecells
writeVideo(outputVideo,imagecell{i});
end
close(outputVideo);
video1 = VideoReader(fullfile(myFolder,'video1.avi'));
mov(video1.NumberOfFrames) = struct('cdata',[],'colormap',[]);
for ii = 1:video1.NumberOfFrames
mov(ii) = im2frame(read(video1,ii));
end
set(gcf,'position', [150 150 video1.Width video1.Height])
set(gca,'units','pixels');
set(gca,'position',[0 0 video1.Width video1.Height])
image(mov(1).cdata,'Parent',gca);
axis off;
movie(mov,1,video1.FrameRate);
Given that there may be too many files to be renamed (padded with zeros) here is a quick function that will do it for you: you just need to provide the directory/folder where the images are stored, the padding (if less 100 files, then padding can be 2; if less than 1000 files, then padding can be 3; etc.), and a common pattern. The code assumes that there is a common pattern in each file (like 'frame' or 'image') that when removed, leaves just the number:
renameFiles(directory,padSize,fileNamePattern)
filePattern = fullfile(directory, '*.jpg') %identify jpg files
jpegFiles = dir(filePattern) %use dir to list jpg files
for k=1:size(jpegFiles)
% get the source file that will be moved/renamed
fileSrc = jpegFiles(k).name;
% get the parts of the file
[path,name,ext] = fileparts(fileSrc);
% where does the pattern fit in the name?
idx = strfind(name,fileNamePattern);
% remove the pattern from the name
if idx==0
% pattern does not exist in file so skip it
continue;
elseif idx==1
% pattern is at the beginning of name so remove it
frameNumberStr = name(length(fileNamePattern)+1:end);
else
% pattern is at the end of name so remove it
frameNumberStr = name(1:idx-1);
end
% get the number of digits
numDigits = length(frameNumberStr);
% create the new file name
paddedName = [fileNamePattern repmat('0',1,padSize-numDigits) frameNumberStr];
fprintf('%s\n',paddedName);
% only move if the destination is different
if strcmp(paddedName,name) ~= 1
% set the destination file
fileDest = fullfile(directory,[paddedName ext]);
% move the file
movefile(fileSrc, fileDest,'f');
end
end
end
An example - if all files have the common pattern of 'frame' and there are less than 1000 files, then run this function as
rename(pwd,3,'frame')
All files that were named as frame1.jpg or frame99.jpg are now named as frame001.jpg and frame099.jpg.
Hope this helps!
If you have the Computer Vision System Toolbox you can use the vision.VideoFileWriter object. You can simply feed images into its step() method one at a time, and they will be written to the video file as video frames. Note that the images must all be the same size and have the same data type.

Subset folder contents Matlab

I have about 1500 images within a folder named 3410001ne => 3809962sw. I need to subset about 470 of these files to process with Matlab code. Below is the section of code prior to my for loop which lists all of the files in a folder:
workingdir = 'Z:\project\code\';
datadir = 'Z:\project\input\area1\';
outputdir = 'Z:\project\output\area1\';
cd(workingdir) %points matlab to directory containing code
files = dir(fullfile(datadir, '*.tif'))
fileIndex = find(~[files.isdir]);
for i = 1:length(fileIndex)
fileName = files(fileIndex(i)).name;
Files also have ordinal directions attached (e.g. 3410001ne, 3410001nw), however, not all directions are associated with each root. How can I subset the folder contents to include 470 of 1500 files ranging from 3609902sw => 3610032sw? Is there a command where you can point Matlab to a range of files in a folder, rather than the entire folder? Thanks in advance.
Consider the following:
%# generate all possible file names you want to include
ordinalDirections = {'n','s','e','w','ne','se','sw','nw'};
includeRange = 3609902:3610032;
s = cellfun(#(d) cellstr(num2str(includeRange(:),['%d' d])), ...
ordinalDirections, 'UniformOutput',false);
s = sort(vertcat(s{:}));
%# get image filenames from directory
files = dir(fullfile(datadir, '*.tif'));
files = {files.name};
%# keep only subset of the files matching the above
files = files(ismember(files,s));
%# process selected files
for i=1:numel(files)
fname = fullfile(datadir,files{i});
img = imread(fname);
end
Something like this maybe could work.
list = dir(datadir,'*.tif'); %get list of files
fileNames = {list.name}; % Make cell array with file names
%Make cell array with the range of wanted files.
wantedNames = arrayfun(#num2str,3609902:3610032,'uniformoutput',0);
%Loop through the list with filenames and compare to wantedNames.
for i=1:length(fileNames)
% Each row in idx will be as long as wantedNames. The cells will be empty if
% fileNames{i} is unmatched and 1 if match.
idx(i,:) = regexp(fileNames{i},wantedNames);
end
idx = ~(cellfun('isempty',idx)); %look for unempty cells.
idx = logical(sum(,2)); %Sum each row