Research On Target Recognition Technology Based On Infrared Hyperspectral Imaging System

Infrared radiation is an electromagnetic wave band which is widely used in today's society and has important applications in military,industrial and medical fields.The infrared detection system,due to its passive detection,has the advantages of strong confidentiality,concealment,not easy to be found and not easy to be disturbed.Target recognition is also the focus of this field.With the emergence of infrared spectral imaging system,infrared detection has developed rapidly in the field of target recognition.It is an advanced technology in this field.In this paper,the detection and recognition of engine flame are mainly carried out by infrared image processing and feature spectrum matching.The image processing part is mainly distinguished by the shape of flame and the difference between the background,while the spectral recognition part is distinguished by the extraction and matching of feature spectrum.This paper firstly analyzes the infrared radiation characteristics of engine flame,studies the radiation theory of infrared technology and the transmission characteristics of infrared in the atmosphere,and verifies that the working band of the infrared spectral imaging system used is practical.The characteristic spectrum of the target is obtained through the simulated infrared target radiation information,which provides a basic template for the subsequent target spectrum matching.Secondly,the performance of infrared spectral imaging system is studied.Based on the system's response curve,noise equivalent power,SNR and other aspects,this paper calculates the NETD value of the short-wave infrared spectral imaging system,and establishes the mathematical model of the system's operating distance through the NETD value,so as to obtain the maximum detection distance of the system against the engine flame.Finally,the infrared target detection,using the image processing and spectral recognition of the fusion of the new method.First applies the improved top-hat transform infrared dim-small target image,similar to the target characteristics of the pixel in the image enhancement,and remove the interference of background on the target,the clouds get some suspected targets enhanced as a result,and combining the threshold segmentation method to enhance to filter results,eliminate most do not conform to the small and weak infrared target characteristics of the background noise points.Then,the spectral matching algorithm is used to carry out weighted matching through three measures to remove the interference of interferers,noises and other factors in image processing,and the filtered target recognition results are obtained.Experimental results show that the algorithm proposed in this paper has better detection effect than traditional small and weak targets detection.