【正文】
??????????? 致謝 ?????????????????????? 參考文獻 ?????? ???????????????? 附錄 ?????????????????????? 數(shù)字圖像中的條帶噪聲方向檢測研究 學 生:徐 冰 指導教師:黃 瑤 (三峽大學 計算機與信息學院) 摘要: 圖像在收集、存儲和傳輸過程中經(jīng)常會受到各類噪聲的干擾 , 主要分為系統(tǒng)噪聲 ( 周期性噪聲 )和隨機噪聲兩種 , 其中條帶噪聲是最常見的系統(tǒng)噪聲。條帶噪聲是圖像中具有一定周期性、方向性且呈條帶狀分布的一種特殊噪聲 , 主要是在采集圖像的過程中 , 由于光電傳感器 —— 線陣 CCD 器件反復掃描的物成像 ,受掃描探測元正反掃 描響應差異、傳感器機械運動和溫度變化等影響造成的。由于條帶噪聲的出現(xiàn)掩蓋了圖像中真正的有用信息 , 使圖像的質(zhì)量下降 , 給圖像識別和判讀造成了不利的影響。而 方向性的有效檢測是去除條帶噪聲干擾的關(guān)鍵所在, 因此 , 在圖像識別和判讀之前 ,利用小波變換對信號奇異點檢測的優(yōu)越性 , 檢測出條帶噪聲在小波系數(shù)中的周期性分布規(guī)律, 根據(jù)條帶噪聲產(chǎn)生的機理, 對條帶噪聲進行有效的方向性檢測,并驗證檢測的準確性,分析誤差產(chǎn)生的原因,然后消除該噪聲。處理后圖像目視效果良好 , 功率譜基本保留了原圖像功率譜的變化趨勢 , 表明在有效去除條帶 噪聲的同時 , 較好地保留了圖像中的有用信息。 關(guān)鍵字:條帶噪聲;周期性;方向性檢測;小波變換;功率譜 Abstract: Images in the collection, storage, and transmission process will often affected by all kinds of noise interference, mainly divided into the system noise, periodic noise and random noise, the stripe noise is the most mon system noise. Stripe noise is in the image has certain periodicity, directional and zonal distribution of a special noise, mainly in the process of collecting image, because of the photoelectric sensor, linear CCD device repeatedly scan imaging, by scanning probe yuan positive and negative scan response differences, sensors, mechanical movement and the temperature change, etc. Due to the presence of the stripe noise to cover up the real useful information in the image, the image quality drop, the adverse effects caused image recognition and directional effective detection is the key to remove stripe noise, therefore, before the image recognition and interpretation, using the advantages of wavelet transform to the signal singularity detection, detect the periodic distribution of strip noise in wavelet coefficient, according to the mechanism of strip noise, for detecting the direction of the stripe noise effectively, and to verify the exactness of the testing, analysis of the causes of error, and then eliminate the processing the image visual effect is good, basic retain the original image power spectrum trend of the power spectrum showed that the effective removal of stripe noise at the same time, better retain the useful information in the image. Keywords: Stripe noise ; Periodically; Directional detection; Wavelet transform;Power spectrum 前言 衛(wèi)星遙感成像是對地觀測的重要手段之一,在國家安全、國民經(jīng)濟、科學研 究和人民生活等方面有重要作用。衛(wèi)星遙感利用地物目標的輻射特性以及對電磁 波的反射特性,實現(xiàn)對地表信息的間接獲取。遙感衛(wèi)星圖像需要真實、客觀的反 映地物特征,然而,由于星載成像系統(tǒng)處于復雜的電磁環(huán)境,衛(wèi)星內(nèi)部電路系統(tǒng) 眾多, CCD 器件的工藝 缺陷等,在獲取和傳輸圖像的過程中會引入大量噪聲,嚴重降低對地觀測信息獲取的圖像質(zhì)量 ,給后期的圖像處理和人眼視覺判讀帶來障礙。因此,對衛(wèi)星成像系統(tǒng)獲得的遙感圖像進行數(shù)據(jù)預處理,降低成像系統(tǒng)中引入的各種噪聲,提高遙感圖像的質(zhì)量和準確度,為后期壓縮編碼、數(shù)據(jù)傳輸、判讀解譯、目標識別等工作能夠順利進行并取得良好結(jié)果提供前提和保證,是獲取遙感圖像后需要完成的首要工作。在星載成像系統(tǒng)獲取圖像的過程中,實際系統(tǒng)受到的噪聲干擾多種多樣,不同的環(huán)境干擾和系統(tǒng)缺陷會對遙感圖像引入分布和