RETRACTED: Algorithm application based on the infrared image in unmanned ship target image recognition

Abstract

Infrared target ship detection is an essential application in the field of military defense and intelligent early warning. Considering the characteristics of intrusion targets and the difficulty of inspection, classification based on Neural Networks (NN) has been proposed for infrared intrusion target detection. This NN calculation joins static objective example investigation with dynamic, diverse connection discovery to remove infrared components at various levels. Among these, infrared boat picture examination can be successfully used to distinguish the back element designs in the objective library. The movement outline contrast strategy, as a camouflage, provides the integrity of the target area. To improve, it can detect the motion area of the image, to combine the benefits of these two methods, an extended convolutional neural network has been designed to fuse and extend the feature images obtained by these two methods. Network expansion modules enhance filter targets and provide background suppression for infrared images. Experimental results contrasted with customary infrared objective discovery techniques. The proposed strategy can recognize infrared interruption targets all the more precisely and smother clamor all the more viably out of sight based on FPGA Xilinx software.