BAYER PATTERN BASED CFA ZOOMING / CFA INTERPOLATION FRAMEWORK (ThuAmOR4)

Author(s) : Rastislav Lukac (University of Toronto, Canada) Karl Martin (University of Toronto, Canada) Konstantinos Plataniotis (University of Toronto, Canada) Bogdan Smolka (Silesian University of Technology, Poland) Anastasios Venetsanopoulos (University of Toronto, Canada)

Abstract : A unified framework for Bayer pattern based single-sensor imaging devices is introduced. Operating on the Bayer color filter array (CFA) data, the method performs CFA image zooming and full color image reconstruction in a cost-effective way making the system practical for hardware implementation. The high-level system components, namely CFA zooming, CFA interpolation and CFA based correction step utilize a color-ratio model and an edge-sensing mechanism to produce naturally colored and sharp, enlarged output. Simulation studies presented here indicate that the new method produces excellent results and outperforms other approaches in terms of both objective and subjective evaluation measures.

Menu