Design and Implementation of Computationally Efficient Image Compressor for Wireless Capsule Endoscopy
Autor: | Shouri Chatterjee, Basabi Bhaumik, Kinde Anlay Fante |
---|---|
Rok vydání: | 2015 |
Předmět: |
Computer science
Color image Applied Mathematics ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION 020207 software engineering Data compression ratio Data_CODINGANDINFORMATIONTHEORY 02 engineering and technology computer.file_format Color space Frame rate Signal Processing 0202 electrical engineering electronic engineering information engineering Electronic engineering Discrete cosine transform 020201 artificial intelligence & image processing Pulse-code modulation Encoder Algorithm computer Image compression |
Zdroj: | Circuits, Systems, and Signal Processing. 35:1677-1703 |
ISSN: | 1531-5878 0278-081X |
Popis: | An image compressor inside wireless capsule endoscope should have low power consumption, small silicon area, high compression rate and high reconstructed image quality. Simple and efficient image compression scheme, consisting of reversible color space transformation, quantization, subsampling, differential pulse code modulation (DPCM) and Golomb---Rice encoding, is presented in this paper. To optimize these methods and combine them optimally, the unique properties of human gastrointestinal tract image are exploited. Computationally simple and suitable color spaces for efficient compression of gastrointestinal tract images are proposed. Quantization and subsampling methods are optimally combined. A hardware-efficient, locally adaptive, Golomb---Rice entropy encoder is employed. The proposed image compression scheme gives an average compression rate of 90.35 % and peak signal-to-noise ratio of 40.66 dB. ASIC has been fabricated on UMC130nm CMOS process using Faraday high-speed standard cell library. The core of the chip occupies 0.018 mm$$^2$$2 and consumes 35 $$\upmu {\text {W}}$$μW power. The experiment was performed at 2 frames per second on a $$256\times 256$$256?256 color image. The power consumption is further reduced from 35 to 9.66 $$\upmu $$μW by implementing the proposed image compression scheme using Faraday low-leakage standard cell library on UMC130nm process. As compared to the existing DPCM-based implementations, our realization achieves a significantly higher compression rate for similar area and power consumption. We achieve almost as high compression rate as can be achieved with existing DCT-based image compression methods, but with an order of reduced area and power consumption. |
Databáze: | OpenAIRE |
Externí odkaz: |