Effects of the retina-inspired light intensity encoding on color discrimination performance

TL;DR

The paper tackles illumination-invariant color discrimination by comparing light-intensity encoding schemes in a center/surround retinex (C/S) framework. It formalizes a logarithmic encoder and a Naka-Rushton (NR) encoder, the latter with $N = V_m \frac{I^n}{I^n + I_h^n}$ and $n=1$, and evaluates their effect on color constancy across HSV and double-opponent color representations. Through a hardware-based, multi-illumination setup, the study shows that the NR-C/S retinex paired with the double-opponent color plane yields superior discrimination for hue, saturation, and brightness compared with the logarithmic encoder and Gray World baseline. These findings highlight the importance of both intensity encoding and color-space representation for robust, illumination-independent color perception in robotic vision, and suggest practical guidelines for adaptive encoding and color-plane selection.

Abstract

Color is an important source of information for visual functions such as object recognition, but it is greatly affected by the color of illumination. The ability to perceive the color of a visual target independent of illumination color is called color constancy (CC), and is an important feature for vision systems that use color information. In this study, we investigated the effects of the light intensity encoding function on the performance of CC of the center/surround (C/S) retinex model, which is a well-known model inspired by CC of the visual nervous system. The functions used to encode light intensity are the logarithmic function used in the original C/S retinex model and the Naka-Rushton (N-R) function, which is a model of retinal photoreceptor response. Color-variable LEDs were used to illuminate visual targets with various lighting colors, and color information computed by each model was used to evaluate the degree to which the color of visual targets illuminated with different lighting colors could be discriminated. Color information was represented using the HSV color space and a color plane based on the classical opponent color theory. The results showed that the combination of the N-R function and the double opponent color plane representation provided superior discrimination performance.

Figures (14)

• Figure 1: Processing flow diagram of the center/surround retinex model for a single channel.
• Figure 2: Input-output relationship of logarithmic transformation for various values of parameter $\gamma$.
• Figure 3: Processing flow diagram of the center/surround retinex model that uses Naka-Rushton function for a single channel.
• Figure 4: Input-output relationship for the logarithmic function and the N-R function.
• Figure 5: Images in the process of computation of the N-R C/S retinex model in the R channel. (a) Output image of the N-R function indicated by $N$ in Fig. \ref{['fig:flow_NRRetinex']}. (b)(c) Gaussian filtered images indicated by $F_{1}$ and $F_{2}$ in Fig. \ref{['fig:flow_NRRetinex']}, respectively. (d) Output image of this channel indicated by $X$ in Fig. \ref{['fig:flow_NRRetinex']}.