Linear demosaicing inspired by the human visual system

Abstract

Abstract-There is an analogy between single-chip color cameras and the human visual system in that these two systems acquire only one limited wavelength sensitivity band per spatial location. We have exploited this analogy, defining a model that characterizes a one-color per spatial position image as a coding into luminance and chrominance of the corresponding three-colors per spatial position image. Luminance is defined with full spatial resolution while chrominance contains sub-sampled opponent colors. Moreover, luminance and chrominance follow a particular arrangement in the Fourier domain, allowing for demosaicing by spatial frequency filtering. This model shows that visual artifacts after demosaicing are due to aliasing between luminance and chrominance and could be solved using a pre-processing filter. This approach also gives new insights for the representation of single-color pe r spatial location images and enables formal and controllable procedures to design demosaicing algorithms that perform well compared to concurrent approaches, as demonstrated by experiments.

A New Color Filter Array With Optimal Properties for Noiseless and Noisy Color Image Acquisition

“…The proposed basis is arbitrary but convenient to describe the spectral properties of the Bayer CFA [11]. Also, the components of natural images can be considered as statistically independent in this basis, in first approximation [21].…”

“…Hirakawa et al [2] work with the G, R − G, B − G basis, which does not decorrelate the image channels at best; also, the chrominance channels in this basis don't have minimum bandwidth, so that the analysis of aliasing is biased. Alleysson et al showed that the mosaicked image v can be interpreted, in the Fourier domain, as the sum of the luminance and chrominance components of the color reference image im, moved at different locations of the frequency plane [11]. We can extend this characterization to every CFA, by writing cfa as the sum of its Fourier components:…”

“…The proposed method for our new CFA is based on the same principles, but the algorithm is even simpler because we don't have to deal with the redundancy of the red-blue chrominance information, a characteristic of the Bayer CFA [11].…”

“…1 (a), is the most popular and dominates the consumer market. There is a vast literature dealing with the best way to reduce aliasing artifacts during the demosaicking process, but these artifacts are inherent to the spectral characteristics of the Bayer CFA [11]. Kodak patented new CFAs containing transparent (panchromatic) filters, in addition to R, G, B filters [3], [12].…”

“…Based on previous work characterizing the spectral properties of the Bayer CFA [11] and showing that the mosaicked image actually consists in the superposition of modulated signals encoding the color information, they proposed to design a CFA so that these signals tile the frequency plane with minimum overlap. This paradigm is quite general, however it leaves open questions about the choice of the many parameters of the model.…”