Eli Intriligator and Max Behrendt

Date: Oct 28 2021

TA help acknowledgements: Dylan, Alex, Gustavo, Igor, Tina, Ann-Marie, Miles, Gustavo, Mary-Joy

The main task of this program is to compress a PPM image into the COMP40 Compressed Image Format and to fully decompress an image in the COMP40 Compressed Image Format back into a PPM image. The compression is lossy, however the effects of the compression are not visible on most images and the compressed format only uses approximately one third of the space required to store the input PPM image. We use component video space conversion, quantization of chroma values, discrete cosine transforms, and bitpacking to accomplish these results.

This program uses a series of classes that break down compression and decompression processes. The colorspace class is used to convert 2d arrays of rgb structs into 2d arrays of ypbpr structs, and also do the reverse. The quantize class is used to quantize the pb and pr values that are stored in ypbpr structs, and also reverse quantize them. The dctrans class takes the y values from ypbpr structs and uses dct to convert them to a, b, c, and d values, and is also used to convert these values back into y values. The bitpack class is used to pack the quantized indicies of pb and pr and a, b, c, and d values into single words. It is also used to unpack these values. compress40.c uses all of these classes to either compress or decompress the given input.

We believe we have implemented all features correctly.

Time spent analyzing problems posed in the assignment: 12 hours

Time spent solving the problems after our analysis: 32 hours