Victor Fernandes's profile - activity

Which are your values for the width and height of the original video ?

Hi Michael!

Probably the best option to this problem, is separate the different channels of your videoline (Y,U and V).

Something like this: Mat yuv422_DDR_RAM = {99, bc, 12, bc, 99, bc, ...}

Mat y = {99,12,99,...} Mat u = {bc,bc,...} Mat v = {bc,...}

After this, you can calculate the histogram of channel Y, using the OpenCV functions and the object cv::Mat y. Repeat the process to channel U and channel V.

If you want the original image you must resize cv::Mat u and cv::Mat v to the same values of width and height of cv::Mat y. After the resizing process you can merge the objects Mat y, u and v.

Cheers, Victor

Hi Seungjun,

Do you know what's the values of width, height? Your file have any header, footer, etc...? What's your chroma subsampling, is 4:4:4?

Hi Nate!

I had the same problem, and I created a solution using OpenCV's functions. Do you want know the solution, or this problem is no longer important? In my case, I received raw data (uncompress video, and the data was in an array of char) from multiples camera feeds (this scenario involved channels Y, U and V separated, different values of chroma subsampling and different values of bit rate). When the Bit Depth was equal to 10-bit I had problems with OpenCV. Using cv::Mat( int height, int width, CV_16UC, (unsigned char*) feed_channel_Y) i didn't had a successful results. In my opinion i think if you use the function cv::VideoCapture, you won't get good results.

To get a solution for this problem, I made some conversions on video data. But remember I worked with uncompressed video data, the video didn't have any codec.

Cheers, Victor