Residue number system in the VLSI architecture for image processing algorithms — A review

Abstract

Improving the computational speed in embedded image processing systems has been of great technical significance. The time elapsed for inter-modular carry propagation in binary arithmetic adversely affect the computational speed and power efficiency of such systems. Different from binary arithmetic, the residue number system has the capability of modular arithmetic and it is devoid of the inter-modular carry propagation. This article reviews image processing applications making use of the residue number system based arithmetic.

A novel image scaling algorithm based on wavelet transform and polyphase filtering

“…The former has the advantage of low complexity, but the scaled image is full of blocky and blending artifacts. The latter solves the mosaic problem caused by the former and can achieve a smoothing effect on the pixel values, but there is also an edge blurring effect [8]. Another linear interpolation method is the Bicubic interpolation algorithm [9].…”

“…In addition, there are a series of interpolation algorithms that will be listed in Table 5. in Section 4.1, which include: window functions such as Box [8], Lanczos [10], Kaiser [11], Mitchell [12], Bell-shaped [13], etc.…”