Weak target detection method in UAV visible light images

Abstract

The invention discloses a method for detecting a weak and small target in a visible light image of an unmanned aerial vehicle, which mainly solves the problem in the prior art that the target is difficult to detect due to scaling and rotation in shape information. The implementation steps are: (1) Input a UAV image containing multiple human targets, extract two image blocks from it as target templates, each image block contains a human target; (2) obtain the target template Mean μ, standard deviation σ, and entropy H; (3) Enhance the input image, and segment the enhanced image according to the color information to obtain multiple superpixel blocks; (4) For all superpixel blocks, sequentially Carry out feature extraction and recognition, and complete the preliminary detection of the target: (5) Remove false targets from the image that has completed the preliminary detection of the above target, and obtain the final detection result. The invention effectively improves the recognition accuracy rate of small and weak targets in the drone image, and can be used in the visible light image or video of the drone.

Description

technical field [0001] The invention belongs to the field of image processing, and further relates to a detection method for weak and small targets, which can be used in visible light images or videos of unmanned aerial vehicles. Background technique [0002] Unmanned Aerial Vehicle (UAV) is a powered, controllable, unmanned aerial vehicle that can carry a variety of mission equipment, perform a variety of tasks, and can be reused. With the continuous improvement of the performance of UAVs, and its advantages of small size, flexibility, and difficulty in being discovered, UAVs are used in reconnaissance and patrolling, building surveys, aerial map drawing, and clearing obstacles in dangerous environments. And other military and civilian special fields have shown huge application potential, so it has been generally valued by countries all over the world. Target tracking has always been a very important task for UAVs. [0003] Existing drones mainly use two methods to achiev...

AI Technical Summary

Problems solved by technology

The shape of the target in the computer can be understood as the contour of the target or the area surrounded by the contour. Based on different understandings, many shape recognition methods have been proposed, such as Fourier descriptors, edges, invariant moments, centroids, rectangularity, etc. , but there are still great difficulties in the application in complex practical situations, such as the matching difficulties caused by scaling and rotation

Images