Research on vehicle detection based on visible light and infrared fusion
Another Title
Научная секция «информационные технологии. Big data. Робототехника. Искусственный интеллект»
Bibliographic entry
Research on vehicle detection based on visible light and infrared fusion = Научная секция «информационные технологии. Big data. Робототехника. Искусственный интеллект» / Hua Gong [и др.] // II Китайско-белорусский молодежный конкурс научно-исследовательских и инновационных проектов : сборник материалов конкурса, 20-21 мая 2021 г. / Белорусский национальный технический университет ; Научно-технологический парк БНТУ «Политехник» ; Институт Конфуция по науке и технике БНТУ. – Минск : БНТУ, 2021. – С. 17.
Abstract
Combining the gamma transform with Sobel edge detection method, an image enhancement
method is designed. The improved Mask R-CNN infrared target detection algorithm based on image enhancement network is proposed. This algorithm is introduced into the Mask R-CNN network based on the decrease of learning rate. Aiming at the limitation of image captured by single sensor, a new target detection algorithm based on decision-level fusion is proposed. This algorithm combines visible light detection with infrared detection. Experiments show that the infrared target detection accuracy is improved by 4.48% after the improved algorithm. The target detection results at decision level are 5.49%~33.80% higher than those of single sensor imaging.