Research On Key Technologies For Long Linear Mid-wave And Long-wave IR Image Fusion

As one of the main directions in the development of the next-generation infrareddetection system,muli-band/multi-color infrared detection technology has increasinglycaught the attention and has obtained large-scale development. In this dissertation,key technologies in multi-band/multi-color infrared focal plane detector(IRFPA)imaging fusion system are studied.The research focus on information fusion andgeneration of a more comprehensive scene description by exploiting the differences oftargets under various wave bands so as to improve the performance of IR detectionsystem.The main contents of the research are divided into four topics as follows:(1) Along with the development of the domestic imaging long linear dual-bandIR detector, the dual-band imaging demonstration prototype is also designed anddeveloped. This prototype has resolved many relevant technical problems such asdual-band image acquisition, misplaced splicing, asynchronous real-time transmissionand GUI, which provides necessary practical evidences for the application of newgeneration dual-band IRFPA.(2) Based on the three indispensible preprocessing aspects of dual-band IRimage fusion, including the high dynamic range compression and enhancement, imagedenoising and image registration, three algorithms are proposed. They are dynamicrange compression algorithm based on total variation(TV), denoising algorithm basedon advanced digital TV filter(DTVF),and registration algorithm based on hybridgenetic algorithm. In this dissertation,the effectiveness of the proposed algorithms areproved by relevant comparative experiments.(3) The differences of the target’s radiation characteristics, atmospheretransmission characteristics, transmittance of infrared optical system characteristicsand response of detectors characteristics are analyzed under middle and long waveband. Furthermore, the theoretical basis of the middle and long wave band imagingfusion is explained. In order to make full and effective use of the information insource images,an evaluation-criteria-driven intelligent fusion model isproposed,which can self-adaptively select suitable fusion methods according tothe regional image features.The fusion results show that the model can effectivelyimprove the quality of the fused image and lay a good foundation for the task-orientedimage fusion.(4) The performance of dual-band image fusion algorithm for automatic target detection has been tested by using real IR images of several typical scenarios.According to the target detection effect and the experimental datas,the characteristicsand application scope of the proposed image fusion model on IR detection system isalso analysed.