This invention discloses a low-light enhancement method from the perspective of a
drone, relating to the field of
image processing and enhancement technology. It involves acquiring a sequence of consecutive low-light image frames of a target during
drone flight, performing preprocessing operations including denoising and normalization to form a low-light image set, and using an image
recognition algorithm to estimate motion between adjacent frames. This invention improves image
clarity and
visibility in low-light environments through
dynamic range compression and detail enhancement
processing. The
dynamic range compression
algorithm balances brightness differences in the images, avoiding local overexposure or underexposure, ensuring good visual effects under different lighting conditions. Detail enhancement
processing, through multi-scale gradient fusion and adaptive
sharpening, highlights texture and edge information in the image, improving detail
clarity in low light. This allows the
drone to more clearly identify targets when performing tasks at night or in low-light environments, improving the efficiency and accuracy of task execution.