A Fast 2D 8×8 DCT Algorithm Based on Look-Up Table for Image Compression

Discrete cosine transform (DCT) has been adopted as an essential part of the well-known image/video compression standards. It is a key factor that DCT can be implemented as fast as possible because of its large number of arithmetic operations in the real-time image transmission. A direct 8×8 2D DCT fast algorithm using look-up table (LUT) is introduced in this paper. The new algorithm is based on the conception and symmetry of basic images. With the new algorithm, the number of addition operations for the transform is reduced while multiplying operations for the transform are eliminated. By designing skillfully the structure of look-up table, one can get a group of product data per memory access, so that the number of looking up the table is reduced greatly. By using the symmetry of basic images and studying the ranges of data in computing the transform, the size of look-up table is decreased. As the new algorithm is executed without involving any multiplication, it is attractive in digital image applications for portable devices, where silicon area and power consumption are dominant issues in hardware design. Moreover, the new algorithm can be parallelized easily. In low bit-rate image compression, only the low-frequency DCT coefficients are computed, which will be encoded by adopting the new algorithm so as to reduce the arithmetic operations greatly.