Adaptive enhancement algorithms of low-quality high-bit infrared-image based on multi-feature fusion and spectral mapping
Infrared images are widely used in the field of autonomous driving, which can effectively compensate for the deficiency of visible images under weak imaging conditions. The infrared images, generated in the autonomous driving field, are mostly 14-bit, high gray-scale ones that cannot be directly displayed on an ordinary 8-bit monitor. Therefore, an adaptive infrared image enhancement algorithm based on multi-feature fusion was firstly proposed to adaptively enhance the infrared images with uneven light and dark distribution. Secondly, to avoid physiological perceptual limitations of human vision for insufficient grayscale discrimination, a new pseudo-color enhancement algorithm for high-bit infrared-images based on spectral mapping was proposed. Finally, to improve adaptive level for the above pseudo-color enhancement problem, a priori luminance statistics method was proposed. The experimental results show that the proposed method can adaptively enhance infrared images with uneven light and dark distribution, with more coordinated enhancement effect, which has obvious advantages compared with other methods. The subjective and objective image quality evaluation indexes are better than those of most existing mainstream advanced algorithms, providing reliable technical support for applications in the related fields.