jdthomas / bayer2rgb Goto Github PK

Raw10 video stream is seen as 8 bit.

You are declaring c as 'char' which is an unsigned datatype on ARM linux such as the Raspberry Pi. Which means -1 becomes 255 and your test against -1 fails.

So getopt_long returns an int for a reason, change the type of 'c' to 'int' to avoid the issue on non-x86 platforms.

And yes, the Internet is full of incorrect examples for getopt usage.

Your code has too many errors to build it with G++ ( 6.3.0 or higher on aarch64 )
Specially MIN, MAX, LIM macros not works correctly, and many typecasting erros to char to int.
And what meaning of "*restricted" ?
It should be fixed.
Please test on armhf(32bit) or aarch64(64bit) platforms of Linux.

how long time change one 1280800 pic from Bayer to RGB? and do you know other faster bayer2RGB algorithm ？I know used opencv：read raw first: rawData(rawDataHeight, rawDataWidth, CV_8UC1, cv::Scalar(0)); then call cvtColor(rawImage, rgbImage, cv::COLOR_BayerGB2BGR), but is also need about 7ms for one 1280800 pic

I would like to use the algorithms from bayer2rgb in my project. I have Nearest Neighbour working perfectly, however, HQLinear and Bilinear attach thin black borders to the delayed image. Is there a way to remove the black borders?

Many thanks