![](/img/trans.png)
[英]How to extract frames from yuv 420 video clip and store them as different images, using matlab?
[英]How to extract Y,U, and V components from a given yuv file using Matlab? Each component is used for further pixels level manipulation
大家好。 我目前正在使用YUV文件。 您對如何從yuv視頻中提取y,u,v分量有任何建議嗎? 我找到了一個程序 ,如下所示。 但是我不知道哪一部分是我想要的有效組件。 謝謝。
% function mov = loadFileYuv(fileName, width, height, idxFrame)
function [mov,imgRgb] = loadFileYuv(fileName, width, height, idxFrame)
% load RGB movie [0, 255] from YUV 4:2:0 file
fileId = fopen(fileName, 'r');
subSampleMat = [1, 1; 1, 1];
nrFrame = length(idxFrame);
for f = 1 : 1 : nrFrame
% search fileId position
sizeFrame = 1.5 * width * height;
fseek(fileId, (idxFrame(f) - 1) * sizeFrame, 'bof');
% read Y component
buf = fread(fileId, width * height, 'uchar');
imgYuv(:, :, 1) = reshape(buf, width, height).'; % reshape RESHAPE(X,M,N) returns the M-by-N matrix
%whose elements are taken columnwise from X.
%An error results if X does not have M*N elements
% read U component
buf = fread(fileId, width / 2 * height / 2, 'uchar');
imgYuv(:, :, 2) = kron(reshape(buf, width / 2, height / 2).', subSampleMat); % reshape and upsample
% read V component
buf = fread(fileId, width / 2 * height / 2, 'uchar');
imgYuv(:, :, 3) = kron(reshape(buf, width / 2, height / 2).', subSampleMat); % reshape and upsample
% normalize YUV values
% imgYuv = imgYuv / 255;
% convert YUV to RGB
imgRgb = reshape(convertYuvToRgb(reshape(imgYuv, height * width, 3)), height, width, 3);
% imgRgb = ycbcr2rgb(imgYuv);
%imwrite(imgRgb,'ActualBackground.bmp','bmp');
mov(f) = im2frame(imgRgb);
% mov(f).cdata = uint8(imgRgb);
% mov(f).colormap = [];
% imwrite(imgRgb,'ActualBackground.bmp','bmp');
%figure, imshow(imgRgb);
%name = 'ActualBackground.bmp';
%Image = imread(name, 'bmp');
%figure, imshow(Image);
end
fclose(fileId);
不知道我是否對YUV文件有一些基本的誤解,但是如果您像下面那樣編輯函數,則每個幀中的YUV組件都在一個名為imgYUV
變量中。 請注意,您可能會在此過程中耗盡內存,或者不想一次性加載電影的所有幀。
function imgYUV = loadFileYuv(fileName, width, height, idxFrame)
% load YUV data from YUV 4:2:0 file
fileId = fopen(fileName, 'r');
subSampleMat = [1, 1; 1, 1];
nrFrame = length(idxFrame);
%# preassign imgYUV. In case we can't keep everything in RAM,
%# it is better that the function crashes here, rather than after
%# having wasted time slowly filling up the memory.
%# Since the images are likely to be of class 'uint8',
%# you can save on a lot of memory by initializing
%# imgYUV as zeros(width/2,height/2,3,nrFrame,'uint8');
imgYUV = zeros(width/2 height/2, 3, nrFrame);
for f = 1 : 1 : nrFrame
%# search fileId position
sizeFrame = 1.5 * width * height;
fseek(fileId, (idxFrame(f) - 1) * sizeFrame, 'bof');
%# read Y component
buf = fread(fileId, width * height, 'uchar');
imgYuv(:, :, 1, f) = reshape(buf, width, height).';
%# read U component
buf = fread(fileId, width / 2 * height / 2, 'uchar');
imgYuv(:, :, 2, f) = kron(reshape(buf, width / 2, height / 2).', subSampleMat); % reshape and upsample
% read V component
buf = fread(fileId, width / 2 * height / 2, 'uchar');
imgYuv(:, :, 3, f) = kron(reshape(buf, width / 2, height / 2).', subSampleMat); % reshape and upsample
end
fclose(fileId);
我只是更改函數定義:
function [mov,imgYuv] = loadFileYuv(fileName, width, height, idxFrame)
無需返回imgRgb,因為它與mov(numel(idxFrame)).cdata
。
指定兩個輸出變量,您將在第二個變量中獲得最后一幀的YUV分量。
如果您請求的幀數超過文件包含的幀數,我將更改該功能以檢查緩沖區是否為空,以避免錯誤。
...
status = fseek(fileId, (idxFrame(f) - 1) * sizeFrame, 'bof');
if status == -1
error('Cannot read frame %d',idxFrame(f));
end
% read Y component
buf = fread(fileId, width * height, 'uchar');
if isempty(buf)
error('Cannot read frame %d',idxFrame(f));
end
imgYuv(:, :, 1) = reshape(buf, width, height).'; % reshape
% read U component
buf = fread(fileId, width / 2 * height / 2, 'uchar');
if isempty(buf)
error('Cannot read frame %d',idxFrame(f));
end
imgYuv(:, :, 2) = kron(reshape(buf, width / 2, height / 2).', subSampleMat); % reshape and upsample
% read V component
buf = fread(fileId, width / 2 * height / 2, 'uchar');
if isempty(buf)
error('Cannot read frame %d',idxFrame(f));
end
...
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.