Sie sind auf Seite 1von 1

Project Members: R.

Srinivas 08EC53 Yashank Sakhardande 08EC67 Surakshith M N 08EC76

IMAGE COMPRESSION USING DCT


ABSTRACT We have decided to do project on 'Image Compression'. We have chosen Discrete Cosine Transform technique to proceed with the idea. After DCT, we plan to perform Huffman coding to remove redundancies. Block Diagram Image in optimal Colour space is Sent to dsp Processor from Matlab platform Compressed Image re-sent to matlab

DCT algo

Quantization

Huffman encoding

Key points Optimal Colour Space :- YCbCr colour space spreads visual image information unevenly across each of its three color components. Y has components sensitive to human eye while Cb & Cr has components less sensitive and hence can be discarded. So we plan to use this colour space. DCT algo :- We resize the input image on matlab platform so as to have a fixed length of DCT. Since we have the 'N' fixed, we can store COSINE factors in the memory and proceed with the algorithm. Quantization:- We divide each component in the frequency domain by a constant for that component (using a quantizing matrix), and then rounding to the nearest integer. The higher frequency components are compressed heavily (as the corresponding coefficient in the quantizing matrix is larger in magnitude). Huffman :- General huffman algorithm is used to remove redundant DCT coefficients.

Das könnte Ihnen auch gefallen