Low-power DCT-based compressor for wireless capsule endoscopy

Abstract

Discrete Cosine Transform (DCT) is known as one of the most significant, fundamental, and prevailing transforms in data compression, watermarking, and medical digital image processing applications for secure storing, sending, and transmitting in cyberspace. This paper addresses a low power, area efficient, and low complexity structure of COordinate Rotation Digital Computer (CORDIC)-based DCT for applying to the Wireless Capsule Endoscopy (WCE) application. In this work, a novel structure is developed called low-power Lookahead (LPLA) CORDIC to overcome the problems of power supply restriction and small dimensions while maintaining the image quality in the WCE. The key idea in the proposed structure is that the successive computations of CORDIC algorithm are performed by utilizing the pipeline property of Lookahead (LA) CORDIC. This leads to employ a single common hardware for the two CORDIC outputs. Thus, the number of the adder and shifter blocks required for CORDIC computation decreases in the proposed structure. Moreover, we utilize pipelining, clock gating and data gating techniques which result in further reduction of power consumption. The simulation results by TSMC 0.13 μm technology at 1.2v power supply demonstrate that the proposed LPLA CORDIC-based DCT structure averagely consumes 64% of the power, 43% of the area, and 8% of the power-delay-product (PDP) of the most efficient structures reported in the literature. Furthermore, by applying the proposed method to JPEG compression standard, it is shown that the reconstructed images of the proposed structure have favorable quality and compression ratio, in addition to achieving lower power consumption and area overhead. According to the aforementioned features, the proposed LPLA CORDIC-based DCT structure is suitable for utilizing in the compressor part of wireless endoscopy capsule and other low-power and high-quality systems which are battery-based.