Aiming at the problem of shadow interference in UAV's ground reconnaissance,a color and polarization synergistic target detection method is proposed for anti-shadow interference,based on the influence of two physi...Aiming at the problem of shadow interference in UAV's ground reconnaissance,a color and polarization synergistic target detection method is proposed for anti-shadow interference,based on the influence of two physical characteristics(wavelength and polarization)under different illuminations.A valid fusion strategy is proposed via integrating two separate detection results on color and polarization images.Moreover,a local enhancement and recognition module(LER)is introduced to boost the performance.Based on our built dataset,experimental results show that our method achieves mAPof 87.76%,and12.37%higher than that of color image and 14.68%higher than that of polarization image.展开更多
基金supported by the Key Research and Development Program of the Key R&D Project in Shaanxi Province (Grant No.2021GXLH-01-20)the Open Fund Project of Key Laboratory of Opto-electronic Information Processing,Chinese Academy of Sciences (Grant No.OEIP-O-202001)+2 种基金the China Industry-universityresearch Innovation Fund (Grant No.2021ITA10006)the National Natural Science Foundation of China (Grant No.62105372)Project Pogram of Science and Technology on Micro-system Laboratory (Grant No.6142804231001)。
文摘Aiming at the problem of shadow interference in UAV's ground reconnaissance,a color and polarization synergistic target detection method is proposed for anti-shadow interference,based on the influence of two physical characteristics(wavelength and polarization)under different illuminations.A valid fusion strategy is proposed via integrating two separate detection results on color and polarization images.Moreover,a local enhancement and recognition module(LER)is introduced to boost the performance.Based on our built dataset,experimental results show that our method achieves mAPof 87.76%,and12.37%higher than that of color image and 14.68%higher than that of polarization image.
