Entropy encoder for low-power low-resources high-quality CFA image compression

Abstract

An entropy encoder for high-quality image compression in low-power, low-resources devices like wireless capsule endoscopy (WCE) or wireless camera sensor network (WCSN) is proposed. The proposed entropy encoder is optimized for encoding data resulting from transformation and quantization of images captured with a color filter array (CFA) digital camera. It efficiently exploits existing correlations among color components in CFA images to improve the performance of the image compressor in terms of power consumption, silicon area, and compression ratio. The image compressor based on the new encoder requires only 22% of the memory size needed by the JPEG-LS-based CFA compressor (for an image with 512 columns). Despite this significant reduction in memory size, the average compression ratio in relation to the image quality is superior to other state-of-the-art algorithms of similar computational complexity and memory requirements. The image compressor based on the proposed encoder has been synthesized using a low-cost UMC 180 nm CMOS process as an intellectual property (IP) core. The design requires a very small silicon area (0.54 × 0.50 mm) and achieves excellent energy efficiency (30μJ per a single 512 × 512 image frame and 20 fps). The selected technology for the implementation is not only low-cost but also well suited to implement an RF transmitter in addition to the compressor to provide a cost effective System on Chip (SoC) for resource-constrained devices like WCE or WCSN.