Image Lossy Compression
Date:
Bachelors Thesis Supervised by Professor Luo Zhong;
Main Work
After a deep study of information theory and image, a hybrid coding method based on discrete cosine transform (DCT) and fractal coding is proposed in combination with the knowledge of the human visual system.
After graphical segmentation, DCT transform is used to highlight the features of the image in the frequency domain for the flat sub-blocks with stable numerical changes, and fractal coding is used for the reduced value domain block pool based on the autocorrelation between the encoded images. At the same time, the search matching degree is increased by combining the equidistant transform, brightness adjustment and brightness shift.
Second segmentation can eliminate the block effect and make the restored image more smooth to increase the fidelity. After horizontal and vertical comparison, the improved algorithm based on MATLAB reduces the coding time and improves image restoration quality.
Technology
- Language: MATLAB
- Information Theory
- DCT Transform and Fractal Coding