manipulating luma in YUV color space

Question

I wont to set contrast / brightnes on image witch is in form byte[]. The image is in YCbCr_420 color space (android camera). I am geting luma value this way :

f         


        

Answer 1

The little that I know from this API is that the values for the 3 channels are concatenated in a byte array. So, likewise in Windows working with RGB, I guess here you have the same, I mean, every 3 bytes representing one pixel. So I GUESS, you could access it jumping every 3 positions to access only one single channel (Luma in your case). I just don't know if the luma is represented by the first, second or third byte.

Second, if I understand you just want to change bright/contrast (increase/decrease) is that correct? Because if so, contrast is just multiplication and brightness is addition.

For instance - pseudo code assuming you are working with 8bits channel:

luma[y] = luma[y] * 1.10; //Increases contrast

or you can have a more generic form:

luma[y] = luma[y] + (luma[y] * contrast); //where contrast ranges from -1.0 to 1.0

Similarly you can do the same with brightness:

luma[y] = luma[y] + bright; //Where bright can range from -255 to 255

In both cases you have to be careful with overflows and underflows before you assign the final pixel result to your image.

Answer 2

YUV processing is best demonstrated in the APLEAGLView.m file of this sample app:

https://developer.apple.com/library/ios/samplecode/AVBasicVideoOutput/Listings/AVBasicVideoOutput_APLEAGLView_m.html#//apple_ref/doc/uid/DTS40013109-AVBasicVideoOutput_APLEAGLView_m-DontLinkElementID_6

Proper adjustments to brightness and contrast so that they maintain their relationship in proportion to each other is demonstrated in the Shader.fsh file at:

https://developer.apple.com/library/ios/samplecode/AVBasicVideoOutput/Listings/AVBasicVideoOutput_Shaders_Shader_fsh.html#//apple_ref/doc/uid/DTS40013109-AVBasicVideoOutput_Shaders_Shader_fsh-DontLinkElementID_9

If you need help interpeting the code, I can help. For now, though, using a user-specified luma threshold value, the luma-adjusted Y plane is calculated like this:

Y' = (Y - (16 / 255)) * lumaThreshold;