Color Constancy Using Double-Opponency.

Area of Science:

• Neuroscience
• Computer Vision
• Computational Neuroscience

Background:

• Color constancy is crucial for visual perception, enabling stable color recognition under varying illumination.
• Double-opponent (DO) cells in the human visual system (HVS) are implicated in color constancy.
• Existing models often lack biological plausibility or require extensive parameter tuning.

Purpose of the Study:

• To propose a novel color constancy model, the Double-Opponency based Color Constancy (DOCC) model.
• To mimic the functional pathway from retinal single-opponent (SO) cells to V1 DO cells and higher visual cortex neurons.
• To leverage DO cell response properties for accurate illuminant estimation.

Main Methods:

• Modeled the human visual system's color processing pathway from retina to V1.
• Utilized the observed correlation between DO cell responses and light source color.
• Developed an illuminant estimation method by pooling DO cell responses in LMS space (sum or max).

Main Results:

• The DOCC model demonstrates competitive performance against state-of-the-art methods.
• Evaluations conducted on three standard datasets with rigorous cross-validation.
• The model achieved strong results without dataset-specific parameter fine-tuning.

Conclusions:

• The physiologically inspired DOCC model offers an effective and simpler approach to color constancy.
• The findings support the role of DO cells in robust illuminant estimation.
• The DOCC model presents a promising alternative for computer vision applications requiring accurate color perception.