Low-Complexity DCT Approximations for Biomedical Signal Processing in Big Data
Implantable, miniaturized, and portable biomedical devices possess severe restrictions, specially in terms of computational and power capabilities. Such constraints impose significant challenges on the design of coding standards commonly found in a multitude of biomedical applications related to image and video coding. Popular image coding algorithms operate over the spectral domain according to tailored transform operators. This chapter reviews the literature of discrete cosine transform (DCT) approximations, a class of low-complexity transforms useful in this type of biomedical applications. Several classical and state-of-the-art DCT approximations are cataloged and compared. The main comparison measures reported in the literature are summarized. In addition to enabling processing on devices with limited resources, DCT approximations have the potential to tackling the vast amounts of data available in biomedical applications in general. Specifically, their utilities in biomedical image compression and image registration are demonstrated. It is hoped that this work demonstrates that approximate DCT methods have the potential to significantly contribute to biomedical signal processing and big data problems.