Robust Looming Spatial Localization in Dim Light via Daubechies Wavelet-Fused ON/OFF Pathways

Area of Science:

• Computational neuroscience
• Bionic vision
• Signal processing

Background:

• Existing computational models of Neohelice granulata MLG1 neurons struggle in dim light due to visual signal noise.
• Photon shot noise significantly degrades performance in low-luminance scenarios.

Purpose of the Study:

• To develop a robust computational framework for detecting and localizing looming stimuli in extremely dim light.
• To improve the performance of MLG1 neuron models under low-contrast conditions.

Main Methods:

• Embedding Daubechies wavelet into ON/OFF visual pathways.
• Utilizing ON/OFF mechanisms for parallel signal separation based on luminance changes.
• Implementing multi-scale frequency decomposition for noise suppression and feature extraction.

Main Results:

• The proposed model demonstrates reliable spatial localization of looming stimuli even in extreme low-contrast conditions.
• The framework effectively suppresses high-frequency noise while enhancing low-frequency looming trends.
• Enhanced feature inputs are provided to the MLG1 neuron model.

Conclusions:

• The computational framework offers a robust methodology for bionic vision in extreme dim light environments.
• Integrating Daubechies wavelet with ON/OFF pathways significantly improves performance in noisy, low-light conditions.