Details

Autor(en) / Beteiligte

• Jaiswal, Ankur
• Garg, Bharat
• Kaushal, Vikas
• Sharma, G. K.

Titel

SPAA-Aware 2D Gaussian Smoothing Filter Design Using Efficient Approximation Techniques

Ist Teil von

• 2015 28th International Conference on VLSI Design, 2015, p.333-338

Ort / Verlag

IEEE

Erscheinungsjahr

2015

Quelle

IEEE Electronic Library (IEL)

Beschreibungen/Notizen

• The limited battery lifetime and rapidly increasing functionality of portable multimedia devices demand energy-efficient designs. The filters employed mainly in these devices are based on Gaussian smoothing, which is slow and, severely affects the performance. In this paper, we propose a novel energy-efficient approximate 2D Gaussian smoothing filter (2D-GSF) architecture by exploiting "nearest pixel approximation" and rounding-off Gaussian kernel coefficients. The proposed architecture significantly improves Speed-Power-Area-Accuracy (SPAA) metrics in designing energy-efficient filters. The efficacy of the proposed approximate 2D-GSF is demonstrated on real application such as edge detection. The simulation results show 72%, 79% and 76% reduction in area, power and delay, respectively with acceptable 0.4dB loss in PSNR as compared to the well-known approximate 2D-GSF.