A Robust Bio-inspired Color Detection Algorithm Under Low/High Light and Shadow for UAV Applications

• Viviane S. S. Freitas IME
• Fabio Suim Chagas IME
• Thiago J. M. Baldivieso IME
• Daniel Morais IME
• Marcos P. S. Campanha IME
• Andre Rotava IME
• Paulo F. F. Rosa IME

Resumo

This project presents a bio-inspired color detection algorithm designed to keep good performance in scenarios where the light conditions are variable and challenging. It is based on the theory of opposite colors, studied in the context of human vision, where the ability to detect colors is associated with a neural response to contrast. To validate the algorithm, we have performed three experiments under adverse luminosity conditions in an indoor environment. Then, we compared the processed images analyzing them in two color spaces - YCrCb and CIEXYZ. As a result, we identified that the YCrCb space provided superior responses and allowed the detection of all eight colors chosen for the experiments.

Palavras-chave: autonomous vision, color spaces, unmanned aerial vehicles (UAV)