The multi-quantization process with matrix size reduction is applied to compress images with strip structure light that is commonly used in 3D reconstructions

Abstract

In this paper, a novel method for 2D image compression is proposed and demonstrated through high-quality image reconstruction with compression ratios up to 99%. The proposed algorithm uses multiple divisions to divide an image into two different matrices: the number of division matrix and the reminder matrix. DCT is applied to these matrices to increase high-frequency coefficients. Then, the final coefficient matrices are encoded using Binary Matrix encoding Algorithm. This final new algorithm removes blocks of zeros and indexes them with only a “0”, while other blocks with nonzero coefficients are kept. At the decompression stage, the process starts with inverse Binary Matrix encoding, which returns all zeros at exact locations. The next step is inverse DCT, which is applied to retrieve the original matrices: the Number of Division matrix and the Reminder matrix. Finally, the image is decoded by combining the two retrieved matrices. The experimental results show that our method achieved high compression ratios up to 99% with better perceptual quality of reconstructed images compared to the popular JPEG method.