Hi Deep, Interpolation for 4:2:0 is done as follows: in effect there will be 3 planes namely Y Cb and Cr the reason for interolating is to achieve compression. the Cb & Cr plane is halved in both horzt & vert directions Lets take without 4:2:0 the avg bit rate per pixel Y,Cb&Cr takes 8 bits each in effect bits/pel will be 24 bits (one pel is 24 bits) after 4:2:0 Y plane is not reduced onluy Cb & Cr planes are halved in both directions. for 4 Y values one Cb & one Cr x x o x x ie, x - Y o - cb & cr lets take 4 Y values named Y1,Y2,Y3,Y4 and corresponding Cb & Cr are Cb1,Cb2,Cb3,Cb4 and Cr1,Cr2,Cr3,Cr4 as mentioned earlier for 4 Y values, 1 Cb & 1 Cr this is done as follows Cb = Cb1+Cb2+Cb3+Cb4/4 Cr = Cr1+Cr2+Cr3+Cr4/4 So effective bits/pel after 4:2:0 is acluclated as follows Y1+Y2+Y3+Y4+Cb+Cr = 48 bits Average bits /pel = 48/4 = 12 bits only! we compressed from 24 to 12 bits/pel ! Rgds, -biju -- In imagedsp@y..., raj deep <deepbluu76@y...> wrote: > Hi > CAn anyone tell me how YCbCr(4:2:0) interpolation > > is done in image processing(source code if exists).My > applicatiion needs to dispaly the YCbCr data thats > being generated after decoding.I have tried with a > certain interpolation filter.I am getting some red > patched on the display. > There isnt any problem with Y .So i guess its problem > with the Cr and Cb interpolation. > regards > deep |