In this paper

A three dimensional discrete cosine transform(3D DCT) coder for still color images is presented

which can not only achieve high compression ratio but also ensure good image quality. In the proposed method

the original image is splitted into non-overlapped 8×8 blocks and segments subblocks of the image into four different classes based on local energy magnitude. Then those subblocks with the same class are gathered to form 3D cuboids

and employs 3D DCT to the 3D cuboids. This improves the efficiency of compression.Computer simulation results show that thePSNR(average Peak Signal-Noise-Ratio) of reconstructed images achieve 29dB with high fidelity when the compression ratios are higher than 40. Compared to JPEG

it advances more than 5dB in the same situation. The proposed method achieves higher fidelity of decoded images than by JPEG.