Embedded Files
- Quality
Define quality (q) as an integer from interval <1, 99>
Compute alpha using formula from the lecture, then multiply both Q matrices with alpha
2. Conversion to YCbCr
2. Conversion to YCbCr
Use rgb2ycbcr to convert your input RGB image to YCbrCr color space
3. Image split
3. Image split
Iterate over image by blocks of 8 x 8, processing one block at the time.
4. Downsampling
4. Downsampling
Downsample Cb and Cr channel (channel 2 and 3) - omit every second column (4:2:2 chroma subsampling / or use different method if you wish)
5. DCT and quantization
5. DCT and quantization
On each channel peform dct2 and divide spectrum with corresponding Q matrix (Y channel - use ql, Cb and Cr - use qc), then round the result
Count number of zeros in the rounded result - compare with total number of pixels in image to get the compression ratio.
DO NOT IMPLEMENT: values from this step are encoded and saved to JPEG using Zig-Zag and some form of compression
6. Inverse DCT
6. Inverse DCT
Perform actions in reversed order
multiply with Q matrices
perform idct2
assign result to decompressed image
upsample where necessary (Cb and Cr channel)
Report abuse