發(fā)布時間：2018-07-24 21:41

【摘要】：在當前大氣污染日益嚴重,霧霾天氣多發(fā)情況下,傳統(tǒng)借助光學傳感器成像的遙感系統(tǒng)的使用受到諸多限制,而天氣狀況等無礙極化合成孔徑雷達(Polarimetric Synthetic Aperture Radar,簡稱Pol SAR)全天時、全天候、全極化模式地獲取地物信息,被廣泛運用于軍事及民用研究領域,其發(fā)展受到廣泛關注。Pol SAR系統(tǒng)通過獲得不同極化狀態(tài)下目標的散射信息,實現(xiàn)地物信息的輕松獲取,如目標的物理特性、空間分布等。但是,高分辨率Pol SAR圖像中,目標散射特性更加復雜,同一地物的不同部分可能呈現(xiàn)出不同的散射特性,低分辨率下的解譯技術不滿足當下Pol SAR系統(tǒng)圖像處理要求�；谌祟惖恼J知機理及已有的認知模型,建立高分辨率Pol SAR圖像目標解譯算法。人類對圖像的認知是目前最高水平的圖像解譯機制,主要歸功于三個特點:一是視覺認知過程是由整體到局部的;二是認知過程分層進行,確保運行過程的高效及高精度;三是知識和經(jīng)驗是認知順利進行的先決條件。深入研究人類的圖像認知機制,并將其引入Pol SAR圖像目標解譯算法中,對于高效、智能、準確地進行圖像解譯具有重要意義。因此,本文基于人類認知機制,結合Pol SAR圖像成像機理,建立“視覺認知-邏輯認知-心理認知”的高分辨率Pol SAR圖像層次認知模型,并且先驗知識全程參與,完成Pol SAR圖像目標地物的快速、準確、智能地解譯與識別。論文第一部分詳細介紹了認知科學、Pol SAR圖像處理等研究的國內(nèi)外研究現(xiàn)狀,如圖像分割技術、圖像目標識別技術。第二部分對認知領域及Pol SAR信息提取與目標解譯技術的基本理論與模型進行了系統(tǒng)地闡述與說明。第三部分基于已有的理論知識,建立了基于先驗知識的“視覺-邏輯-心理”認知的Pol SAR圖像層次認知模型,并借助多層次圖像分割、模糊邏輯、神經(jīng)網(wǎng)絡、上下文語義特征等理論,完成了相應的數(shù)學構建及編程實現(xiàn)。第四部分,利用三組不同成像條件下的Pol SAR圖像所獲得的實驗結果進行算法驗證。并且定量的結果分析表明,本文所提算法具有很好地圖像解譯效果,且具有一定的普適性,對Pol SAR圖像解譯研究具有重要的應用價值和意義。
[Abstract]:With the increasing air pollution and the frequent weather in haze, the use of traditional remote sensing system with optical sensor imaging is restricted by many restrictions, while the weather conditions such as weather conditions do not hinder the polarimetric synthetic Aperture Radar (Polarimetric Synthetic Aperture Radar,), which is referred to as Pol SAR), all day long. It is widely used in military and civilian research fields to obtain ground object information in all-weather, all-polarization mode. The development of .Pol SAR system can easily obtain ground object information by obtaining scattering information of target in different polarization state. Such as the physical characteristics of the target, spatial distribution, etc. However, in high-resolution Pol SAR images, the scattering characteristics of targets are more complex, and different parts of the same object may exhibit different scattering characteristics. The interpretation techniques at low resolution do not meet the requirements of current Pol SAR system image processing. Based on human cognitive mechanism and existing cognitive models, a target interpretation algorithm for high resolution Pol SAR images is established. Human cognition of image is the highest image interpretation mechanism at present, which is mainly attributed to three characteristics: one is that the visual cognitive process is from whole to local, the other is that the cognitive process is stratified to ensure the high efficiency and high accuracy of the operation process. Third, knowledge and experience is a prerequisite for the smooth development of cognition. It is of great significance for efficient, intelligent and accurate image interpretation to deeply study human image cognitive mechanism and introduce it into Pol SAR image target interpretation algorithm. Therefore, based on the human cognitive mechanism and the imaging mechanism of Pol SAR images, a high-level cognitive model of "visual cognitive-logical cognitive-psychological cognition" for high-resolution Pol SAR images is established, and prior knowledge is involved in the whole process. Fast, accurate and intelligent interpretation and recognition of target objects in Pol SAR images are completed. The first part of this paper introduces the research status of cognitive science Pol SAR image processing at home and abroad, such as image segmentation technology, image target recognition technology. In the second part, the basic theories and models of cognitive domain and Pol SAR information extraction and target interpretation technology are systematically expounded and explained. In the third part, based on the existing theoretical knowledge, the Pol SAR image hierarchical cognitive model based on transcendental knowledge is established, and with the help of multi-level image segmentation, fuzzy logic, neural network. The theory of contextual semantic features has completed the corresponding mathematical construction and programming implementation. In the fourth part, the experimental results of three groups of Pol SAR images under different imaging conditions are used to verify the algorithm. The quantitative results show that the proposed algorithm has a good map image interpretation effect, and has a certain universality, which has important application value and significance to the research of Pol SAR image interpretation.
【學位授予單位】：哈爾濱工業(yè)大學
【學位級別】：碩士
【學位授予年份】：2015
【分類號】：TN957.52

【參考文獻】

相關期刊論文 前1條

1 肖鵬;徐軍;陳少沖;;紋理特征提取方法[J];電子科技;2010年06期

相關博士學位論文 前1條

1 張臘梅;極化SAR圖像人造目標特征提取與檢測方法研究[D];哈爾濱工業(yè)大學;2010年

相關碩士學位論文 前1條

1 高貴;高分辨率SAR圖像目標特征提取研究[D];國防科學技術大學;2003年

，

本文編號：2142741