www.gusucode.com > ros 工具箱 matlab源码程序 > ros/+robotics/+ros/+msg/+sensor_msgs/+internal/ImageReader.m
classdef ImageReader
%This class is for internal use only. It may be removed in the future.
%ImageReader Class for converting ROS image messages to MATLAB images
% This class contains utility functions for converting several image
% encodings into representations that are usable within MATLAB.
%
% See also: robotics.ros.msg.sensor_msgs.Image
% Copyright 2014 The MathWorks, Inc.
methods (Static)
function img = convertToGray(rawData, width, height, dataType)
%convertToGray Convert the image data to a MATLAB gray-scale image
% The user has to specify information about the element data
% type to correctly interpret the data.
%
% IMG = convertToGray(RAWDATA, WIDTH, HEIGHT, DATATYPE) converts
% the raw image data contained in the ROS image data RAWDATA
% into an image IMG that can be used for further processing in MATLAB.
% IMG will be a gray-scale image. Its size will be HEIGHT x WIDTH.
% DATATYPE is the MATLAB data type for each
% stored pixel value in a channel.
%
% Examples:
% % Convert a raw 8-bit image into MATLAB image
% img = convertToGray(data, 640, 480, 'uint8');
%
% % Convert a raw 16-bit image into MATLAB image
% img = convertToGray(data, 640, 480, 'uint16');
img = robotics.ros.msg.sensor_msgs.internal.ImageReader.convertToImg(...
rawData, width, height, 1, [], dataType);
end
function img = convertToBayer(rawData, width, height, alignment, dataType)
%convertToBayer Convert the image data to a MATLAB bayer image
% If the Image Processing Toolbox is installed, the raw image
% will be de-bayered and returned as an RGB image. Otherwise,
% the image will be returned as a 1-channel image.
%
% IMG = convertToBayer(RAWDATA, WIDTH, HEIGHT, ALIGNMENT, DATATYPE) converts
% the raw image data contained in the ROS image data RAWDATA
% into an image IMG that can be used for further processing in MATLAB.
% Its size will be HEIGHT x WIDTH. If the Image Processing
% Toolbox is installed, the return image will be an RGB image
% of size HEIGHT x WIDTH x 3. DATATYPE is the MATLAB data type for each
% stored pixel value in a channel. ALIGNMENT is the sensor
% element alignment
%
% Example:
% img = convertToBayer(data, 640, 480, 'bggr', 'uint16');
img = robotics.ros.msg.sensor_msgs.internal.ImageReader.convertToImg(...
rawData, width, height, 1, [], dataType);
% Also debayer the image if IPT is installed
if robotics.ros.msg.sensor_msgs.internal.ImageLicense.isIPTInstalled
img = demosaic(img, alignment);
end
end
function [img,alpha] = convertToRGBA(rawData, width, height, numChannels, channelOrder, dataType)
%convertToRGBA Convert the image data to RGB and Alpha
% The user has to specify information about the number of
% channels, channel order, and data type to correctly
% interpret the data.
% Get full image first (height x width x 4)
imgFull = robotics.ros.msg.sensor_msgs.internal.ImageReader.convertToImg(...
rawData, width, height, numChannels, channelOrder, dataType);
% Split it up into components
img = imgFull(:,:,1:3);
alpha = imgFull(:,:,4);
end
function img = convertToRGB(rawData, width, height, numChannels, channelOrder, dataType)
%convertToRGB Convert the image data to a MATLAB RGB image
% The user has to specify information about the number of
% channels, channel order, and data type to correctly
% interpret the data.
%
% IMG = convertToRGB(RAWDATA, WIDTH, HEIGHT, NUMCHANNELS, CHANNELORDER,
% DATATYPE) converts the raw image data contained in the ROS
% image data RAWDATA and converts it into an RGB image
% IMG that can be used for further processing in MATLAB.
% NUMCHANNELS is the number of channels that the image has
% and CHANNELORDER is a 3-vector of integers that specify
% where in the image the R, G, and B channels can be found,
% respectively. DATATYPE is the MATLAB data type for each
% stored pixel value in a channel.
%
% Examples:
% % Convert a raw BGRA 8-bit image into MATLAB image
% img = convertToRGB(data, 640, 480, 4, [3,2,1], 'uint8');
%
% % Convert a raw RGB 16-bit image into MATLAB image
% img = convertToRGB(data, 640, 480, 3, [1,2,3], 'uint16');
img = robotics.ros.msg.sensor_msgs.internal.ImageReader.convertToImg(...
rawData, width, height, numChannels, channelOrder, dataType);
end
function img = convertToImg(rawData, width, height, numChannels, channelOrder, dataType)
%convertToImg Convert raw image data to a MATLAB matrix
% This function is used as a backend for all image conversion
% functions. It is optimized for speed to provide a flexible
% and performant interface to retrieve image data.
%
% IMG = convertToImg(RAWDATA, WIDTH, HEIGHT, NUMCHANNELS,
% CHANNELORDER, DATATYPE) converts the raw image data contained
% in the ROS image data RAWDATA and converts it into an image
% IMG that can be used for further processing in MATLAB. IMG
% will be a HEIGHT x WIDTH x NUMEL(CHANNELORDER) matrix.
% NUMCHANNELS is the number of channels that the image has
% and CHANNELORDER is a vector of integers that specify
% where in the raw stream the output channels can be found.
% DATATYPE is the MATLAB data type for each
% stored pixel value in a channel.
%
% Examples:
% % Convert a raw BGRA 8-bit image into RGB MATLAB image
% img = convertToImg(data, 640, 480, 4, [3,2,1], 'uint8');
%
% % Keep the alpha channel of the BGRA 8-bit image
% % Output img will be 480x640x4
% img = convertToImg(data, 640, 480, 4, [3,2,1,4], 'uint8');
%
% % Convert a raw RGB 16-bit image into MATLAB image
% img = obj.convertToRGB(data, 640, 480, 3, [1,2,3], 'uint16');
% Return if no data was received
if isempty(rawData)
img = rawData;
return;
end
% The output image will be height x width x numOutputChannels
numOutputChannels = numel(channelOrder);
% Cast input data to the expected data type
data = typecast(rawData, dataType);
% Ensure that width, height, and number of channels are
% consistent with raw stored data
expectedElems = width*height*numChannels;
if expectedElems ~= numel(data)
error(message('robotics:ros:image:DataLengthInconsistency', ...
num2str(expectedElems), num2str(width), num2str(height), ...
num2str(numChannels), num2str(numel(data))));
end
% If this is a single-channel image, return right away
if numChannels == 1
img = reshape(data, width, height)';
return;
end
% Construct the data access index. Since the channel ordering
% might be arbitrary, this has to be done in a for loop
indexBase = (1:numChannels:width*height*numChannels)';
for i = 1:numOutputChannels
% Construct it as column-major index to ease later reshape
index(:, channelOrder(i)) = indexBase + i - 1; %#ok<AGROW>
end
% Reshape the data into a width x height x numOutputChannels matrix
img = reshape(data(index), width, height, numOutputChannels);
% Since reshape works column-wise, bring image back to
% row-major order (transpose of 1st and 2nd dimension
% Result will be a height x width x numOutputChannels image matrix
img = permute(img, [2 1 3]);
end
function img = decompressImg(rawData, numChannels)
%decompressImg Take raw data and decompress
try
compressedData = org.jboss.netty.buffer.ChannelBufferInputStream(rawData);
inputStream = javax.imageio.ImageIO.createImageInputStream(compressedData);
imjava = javax.imageio.ImageIO.read(inputStream);
w = imjava.getWidth();
h = imjava.getHeight();
imageData = uint8( imjava.getData.getPixels( 0,0,w,h,[] ) );
img = permute( reshape( imageData, [numChannels w h] ), [3 2 1] );
catch ex
newEx = robotics.ros.internal.ROSException( ...
message('robotics:ros:image:DecompressError'));
throw(newEx.addCustomCause(ex));
end
end
end
end