Does this method work better for a image with smooth changes of colors around edges?
motoole2
@Qwerty There are certainly better compression techniques out there. ;-) But if your chromaticity images contain primarily low-frequency information (lowo-freq. changes in chromaticity across the image), then it can be easily downsampled without loss of information.
elenagong
How is the color deviation decided? If we use some other color deviation, will it work as good as Cb and Cr do?
motoole2
@elenagong I would point you to the wiki page for YCbCr for details on computing each component from RGB values. If you use "some other color deviation", this just means you are employing a different color model. As for your question on whether it would be as good as Cb and Cr, the answer depends a lot on your custom color model and how it will be used.
Does this method work better for a image with smooth changes of colors around edges?
@Qwerty There are certainly better compression techniques out there. ;-) But if your chromaticity images contain primarily low-frequency information (lowo-freq. changes in chromaticity across the image), then it can be easily downsampled without loss of information.
How is the color deviation decided? If we use some other color deviation, will it work as good as Cb and Cr do?
@elenagong I would point you to the wiki page for YCbCr for details on computing each component from RGB values. If you use "some other color deviation", this just means you are employing a different color model. As for your question on whether it would be as good as Cb and Cr, the answer depends a lot on your custom color model and how it will be used.