Armored Vehicle Fusion Detection Auxiliary Driving System

The mobility of armored vehicles determines that their driving condition is usually complex and changeable.The role of the vehicle's driving assist system is to collect information and process it through detectors to assist the driver in observing the surroundings.The traditional military vehicles are equipped with the visual detection auxiliary systems which are mostly based on single detector,usually equipped with a single infrared detector or a single low-light detector.Although the single-source detector can improve the vision of the human eye to a certain extent,but the disadvantage is vulnerable to the environment.In the paper,a set of armored vehicle fusion detection driving assist system is designed.Improving imaging quality through image fusion.Target detection on the fusion image makes the armored vehicle detect target efficient and accurate in a complex environment,it is easy for drivers to observe the environment and detect targets.First of all,comparing the domestic and foreign research status of the armored vehicle driving assist system,and the development of fusion detection and target detection are briefly summarized in the paper,and then the significance of the design of the system and the specific design scheme are explained.Secondly,researching on the fusion imaging algorithm in the system,and optimize the algorithm for the image fusion algorithm.A fast matching algorithm based on FAST feature points is designed.Based on the Laplacian pyramid image fusion algorithm,the paper designed an image fusion function which is based on Domain-related fusion criteria.The image fusion function is implemented on the FPGA hardware platform.Then,the research on multi-source feature is carried out,and a convolutional neural network is designed based on the ideas of residual neural network and feature pyramid network;the fuzzy neural network is used for feature fusion of multi-source images,and the image design after feature fusion designs a new image detection algorithm based on fused images was transplanted to the Jetson TX2 platform,and the GPU deployment was completed through neural network optimization.Finally,after building a system,testing the image fusion and target detection functions in the system.Experiments show that the fused image is more conducive to the human eye observation,and the fusion target detection rate is higher than that of the single target detection algorithm.