Eagle eye midbrain loop return inhibition mechanism-imitated unmanned target drone target detection method

Abstract

The invention discloses an eagle eye midbrain loop return inhibition mechanism-imitated unmanned target drone target detection method, which comprises the following steps of: 1, calibrating a camera,acquiring internal and external parameters of the camera, and correcting image distortion; 2, carrying out the illumination processing of imitating the color constancy of the beasts, and recovering the color information of the image under the normal illumination; 3, preprocessing an eagle midbrain opening and closing corresponding mechanism-imitated image; 4, extracting a salient region based on an eagle visual sensitivity function; and step 5, carrying out target output based on an eagle brain return inhibition mechanism. The method has the advantages that 1, an eagle eye biological image processing mechanism is used for reference, and an eagle midbrain loop return inhibition target detection framework is extracted and applied to small target detection of an unmanned target drone; 2, themethod can adapt to the small target detection problem of the target drone in various environments, the influence of external factors such as illumination and contrast is eliminated, and the algorithmadaptability is improved; and 3, the method is simple in framework, the target detection calculation amount is saved, and the airborne calculation load requirement is greatly reduced.

Description

technical field [0001] The invention relates to a target detection method of an unmanned target drone imitating the return inhibition mechanism of the midbrain circuit of the eagle eye, and belongs to the field of computer vision. Background technique [0002] The target drone is a military aircraft used to simulate air maneuvering targets. It plays a very important role in the development and inspection of air-to-air weapons and the combat training of troops. With the development of materials, electronics, sensors and batteries, drone technology is also constantly developing. Target drone technology can provide advanced experience for various types of equipment including unmanned reconnaissance aircraft, unmanned fighter jets, unmanned underwater vehicles, unmanned surface craft, unmanned intelligent tanks and armored vehicles, military robots, etc. , optoelectronics, nano, micro-electromechanical, computer, communication, new materials, new power and many cutting-edge ach...

AI Technical Summary

Problems solved by technology

The currently popular deep learning method also has applications in small target detection, but because deep learning methods generally require high computing power, the aerial platform cannot meet the demand for computing load

The deep learning method has a large amount of calculation, high demand for hardware, and the model design process is cumbersome. Some technologies lack theoretical support and have a certain dependence on parameter adjustment.

Moreover, the deep learning method is more sensitive to noise and is easily affected by interference and ignores the target information.

Images