红外与激光工程
紅外與激光工程
홍외여격광공정
INFRARED AND LASER ENGINEERING
2014年
8期
2746-2751
,共6页
杨蔚%顾国华%陈钱%曾海芳%徐富元%王长江
楊蔚%顧國華%陳錢%曾海芳%徐富元%王長江
양위%고국화%진전%증해방%서부원%왕장강
红外偏振%Mean-Shift 聚类%DS 证据理论%模式识别
紅外偏振%Mean-Shift 聚類%DS 證據理論%模式識彆
홍외편진%Mean-Shift 취류%DS 증거이론%모식식별
infrared polarization%Mean-Shift clustering%DS evidence theory%pattern recognition
红外偏振成像探测通过对目标辐射和反射偏振态的探测,针对传统光学无法解决的问题,在目标检测方面取得高精度的结果,特别是在军事探测中,能够快速地将混杂在自然背景下的人造目标检测出来,以增强对目标的识别。偏振探测中所依据的强度、偏振度及偏振角信息反映出的不同物理特性,具有很强的冗余性和互补性。针对该特性,提出一种红外偏振图像的目标检测方法:首先使用Mean-Shift 算法对红外图像和偏振度图像进行聚类处理;然后利用 DS 证据理论将聚类后的红外图像和偏振度图像中的物体信息充分结合,以区分目标与背景,达到目标检测的目的;最后通过仿真实验图像与小波融合图像结果的对比表明该算法的优势。
紅外偏振成像探測通過對目標輻射和反射偏振態的探測,針對傳統光學無法解決的問題,在目標檢測方麵取得高精度的結果,特彆是在軍事探測中,能夠快速地將混雜在自然揹景下的人造目標檢測齣來,以增彊對目標的識彆。偏振探測中所依據的彊度、偏振度及偏振角信息反映齣的不同物理特性,具有很彊的冗餘性和互補性。針對該特性,提齣一種紅外偏振圖像的目標檢測方法:首先使用Mean-Shift 算法對紅外圖像和偏振度圖像進行聚類處理;然後利用 DS 證據理論將聚類後的紅外圖像和偏振度圖像中的物體信息充分結閤,以區分目標與揹景,達到目標檢測的目的;最後通過倣真實驗圖像與小波融閤圖像結果的對比錶明該算法的優勢。
홍외편진성상탐측통과대목표복사화반사편진태적탐측,침대전통광학무법해결적문제,재목표검측방면취득고정도적결과,특별시재군사탐측중,능구쾌속지장혼잡재자연배경하적인조목표검측출래,이증강대목표적식별。편진탐측중소의거적강도、편진도급편진각신식반영출적불동물리특성,구유흔강적용여성화호보성。침대해특성,제출일충홍외편진도상적목표검측방법:수선사용Mean-Shift 산법대홍외도상화편진도도상진행취류처리;연후이용 DS 증거이론장취류후적홍외도상화편진도도상중적물체신식충분결합,이구분목표여배경,체도목표검측적목적;최후통과방진실험도상여소파융합도상결과적대비표명해산법적우세。
Infrared polarization imaging measurement can be used to obtain the polarization state in target radioactive and reflective detection. With this method, the problems that traditional photometry cannot solve can be settled, and high-precision results were obtained. Particularly, in the field of military detection, polarization measurement can be introduced to distinguish artificial target quickly from the natural background, and performed well in target identification. The intensity, degree and angle of polarization that used in polarization measurement can reflect different physical properties, and along with them, redundancy and complementarity are revealed seriously. In allusion to the character, a target detection method for infrared polarization image was proposed: firstly, Mean-Shift algorithm was employed to aggregate the infrared and polarized image. Secondly, the object information of clustered infrared and polarized image according to DC evidence theory were combined. In this case, the target can be distinguished from the background, so as to achieve the purpose of target detection. Finally,comparison of the consequence of experimental image and wavelet fusion image to demonstrate the superiority of the presented algorithm.