A novel approach towards high-performance image compression using multilevel wavelet transformation for heterogeneous datasets

Abstract

Multimedia information typically requires large storage capacity and transmission rate. Thus, multimedia data, which exist in various forms, such as images, audios, videos, and graphics, require compression prior to transmission because of their large size. Furthermore, they cannot be stored directly into storage devices with less capacity. Hence, image compression is the best solution for decreasing the size of images considerably while sustaining their quality. Moreover, image redundancy must be avoided for reducing their size. In this regard, compression methods aid in reducing image redundancy. A majority of existing image-compression techniques are well-suited for grayscale image compression, but only few methods have been reported to be effective for color image compression. However, current color-image compression algorithms result in high information loss and increased computational cost. Hence, reducing image size without degrading its quality, reducing image redundancy, and effectively compressing color images with reduced information loss are critical issues that require immediate attention. To address these issues, we proposed a novel image compression algorithm referred to as the Bayer wavelet transform (BWT) algorithm. The features of this algorithm include directly compressing RGB images without converting them to grayscale, preprocessing module largely suppressing noise from input images, and the proposed Bayer encoder compressing heterogeneous data to reduce storage requirements. This method was used for analyzing iris medical image and commercial two-dimensional image datasets. Few notable inferences concerning both the datasets included the reduction in the bit depth of images from conventional 24 bits to 8 bits during image compression, recovery of the compressed images directly in their original form without transformation to grayscale, and compression of input images in every dimension to preserve their edges. These results revealed that the BWT algorithm outperformed existing techniques in terms of accuracy, compression ratio, mean square error, peak signal-to-noise ratio, contrast improvement index, and computational time complexity. Additionally, by obtaining the highest compression ratio, the compression and decompression processes were completed in 0.0190 ms and 0.0189 ms, respectively, which validated the applicability of this algorithm in high-performance image-compression systems, particularly in telemedicine. Ultimately, the memory requirements for this algorithm were reduced to a ratio of 3:1.