本文選題：遙感圖像 + 圖像增強(qiáng)��； 參考：《新疆大學(xué)》2017年碩士論文

【摘要】：遙感圖像作為人們對(duì)地球進(jìn)行研究和監(jiān)測(cè)的一個(gè)重要的依據(jù),起到十分重要的作用。遙感已經(jīng)從軍事領(lǐng)域轉(zhuǎn)向了民用方面,由此使其發(fā)展的更加迅速。但是遙感圖像在獲取和傳輸?shù)倪^(guò)程中,受到很多因素的影響,如傳感器、大氣等,使得到的遙感圖像變得失真、模糊、對(duì)比度低等。為了便于研究人員對(duì)遙感圖像的識(shí)別,就必須對(duì)遙感圖像進(jìn)行處理。圖像增強(qiáng)是一種圖像處理方法,主要是針對(duì)圖像中的一些有用信息進(jìn)行突出或強(qiáng)化。因此,圖像增強(qiáng)對(duì)遙感圖像的識(shí)別是必不可少的。本文針對(duì)遙感圖像存在的低對(duì)比度、低信噪比、邊緣保持較弱、細(xì)節(jié)丟失等問(wèn)題,提出了兩種新的圖像增強(qiáng)算法。其中一種是基于在NSST域的自適應(yīng)閾值和引導(dǎo)濾波相結(jié)合的遙感圖像增強(qiáng)算法。引導(dǎo)濾波是一種圖像濾波算法,具有良好的平滑能力的同時(shí)還能對(duì)圖像邊緣梯度能很好的保持,得到了研究人員的關(guān)注。鑒于這些特性,本文將結(jié)合引導(dǎo)濾波對(duì)圖像的細(xì)節(jié)和邊緣部分進(jìn)行增強(qiáng)。首先,該算法通過(guò)對(duì)待處理的圖像進(jìn)行NSST分解,將圖像分解成為一個(gè)低頻部分和若干個(gè)高頻部分。然后,采用線性變換對(duì)低頻部分進(jìn)行線性拉伸,目的在于對(duì)比度的改變;高頻部分,進(jìn)行抑制噪聲處理,將采用自適應(yīng)閾值法,其次再進(jìn)行引導(dǎo)濾波增強(qiáng)圖像的細(xì)節(jié)部分和邊緣梯度。最后,將處理后低頻和高頻部分進(jìn)行重構(gòu)處理,得到增強(qiáng)后的圖像。通過(guò)實(shí)驗(yàn)表明,該算法對(duì)遙感圖像的視覺(jué)效果得到了改善,客觀指標(biāo)上與對(duì)比算法相比,信息熵、峰值信噪比和結(jié)構(gòu)相似度有了一定的提升。本文另外一種是基于NSST域的直方圖均衡和引導(dǎo)濾波相結(jié)合的遙感圖像增強(qiáng)算法。直方圖均衡是一種經(jīng)典的用來(lái)提高對(duì)比度的算法,本文用它來(lái)對(duì)圖像進(jìn)行預(yù)處理,提高圖像整體的對(duì)比度。經(jīng)過(guò)NSST分解后的低頻部分的處理與上一種算法一樣采用線性變換的方式,高頻部分的去噪處理采用閾值去噪的方法,對(duì)于圖像的細(xì)節(jié)部分和邊緣,還是采用引導(dǎo)濾波的方法。實(shí)驗(yàn)表明,與對(duì)比算法相比較,該算法明顯地提升了圖像的對(duì)比度,增強(qiáng)了圖像的細(xì)節(jié)和邊緣梯度能力。
[Abstract]:Remote sensing image plays an important role as an important basis for people to study and monitor the earth. Remote sensing has shifted from military to civilian, thus making its development more rapid. However, in the process of obtaining and transmitting remote sensing image, it is affected by many factors, such as sensor, atmosphere and so on, which make the remote sensing image become distorted, blurred and low contrast. In order to facilitate the recognition of remote sensing images, remote sensing images must be processed. Image enhancement is an image processing method, which is mainly used to highlight or enhance some useful information in the image. Therefore, image enhancement is essential for remote sensing image recognition. In this paper, two new image enhancement algorithms are proposed to solve the problems of low contrast, low signal-to-noise ratio (SNR), weak edge retention and detail loss in remote sensing images. One is a remote sensing image enhancement algorithm based on adaptive threshold and guided filtering in NSST domain. The guided filter is a kind of image filtering algorithm, which has good smoothing ability and can keep the edge gradient of the image well, which has been paid attention to by researchers. In view of these features, this paper will enhance the details and edges of the image in combination with bootstrap filtering. Firstly, the algorithm decomposes the image into a low frequency part and several high frequency parts by NSST decomposition. Then, the low frequency part is stretched linearly by linear transformation to change the contrast, and the high frequency part, which is used to suppress noise, will adopt the adaptive threshold method. Secondly, the detail part and edge gradient of the image are enhanced by guided filtering. Finally, the processed low frequency and high frequency parts are reconstructed to get the enhanced image. The experiments show that the visual effect of the algorithm is improved, and the information entropy, peak signal-to-noise ratio and structural similarity are improved compared with the contrast algorithm. The other one is a remote sensing image enhancement algorithm based on histogram equalization and guided filtering in NSST domain. Histogram equalization is a classical algorithm to improve the contrast. In this paper, we use it to preprocess the image to improve the overall contrast of the image. The processing of the low-frequency part after NSST decomposition is the same as that of the previous algorithm. The high-frequency part is de-noised by the threshold denoising method, and the image details and edges are still processed by the guided filtering method. The experimental results show that compared with the contrast algorithm, the algorithm improves the contrast of the image, and enhances the ability of image detail and edge gradient.
【學(xué)位授予單位】：新疆大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP751

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 呂篤良;賈振紅;楊杰;Nikola KASABOV;;基于非下采樣剪切波變換與引導(dǎo)濾波結(jié)合的遙感圖像增強(qiáng)[J];計(jì)算機(jī)應(yīng)用;2016年10期

2 王焱;關(guān)南楠;劉海濤;;改進(jìn)的多尺度Retinex井下圖像增強(qiáng)算法[J];遼寧工程技術(shù)大學(xué)學(xué)報(bào)(自然科學(xué)版);2016年04期

3 劉燕妮;張貴倉(cāng);安靜;;基于數(shù)學(xué)形態(tài)學(xué)的雙直方圖均衡化圖像增強(qiáng)算法[J];計(jì)算機(jī)工程;2016年01期

4 陶飛翔;吳一全;;非下采樣Shearlet變換與參數(shù)化對(duì)數(shù)圖像處理相結(jié)合的遙感圖像增強(qiáng)[J];測(cè)繪學(xué)報(bào);2015年08期

5 吳一全;孟天亮;吳詩(shī)Zs;;基于NSST和人眼感知保真約束的圖像自適應(yīng)增強(qiáng)[J];光電子·激光;2015年05期

6 吳一全;孟天亮;吳詩(shī)Zs;;人工蜂群優(yōu)化的非下采樣Shearlet域引導(dǎo)濾波圖像增強(qiáng)[J];西安交通大學(xué)學(xué)報(bào);2015年06期

7 舒婷;劉耀峰;鄧波;譚亞鵬;陳炳權(quán);;基于子帶分解的MSR霧天圖像增強(qiáng)算法[J];吉首大學(xué)學(xué)報(bào)(自然科學(xué)版);2015年01期

8 王靜靜;賈振紅;覃錫忠;楊杰;Nikola KASABOV;;基于Shearlet變換和多尺度Retinex的遙感圖像增強(qiáng)算法[J];計(jì)算機(jī)應(yīng)用;2015年01期

9 田小平;李曉敏;;基于小波變換的遙感圖像增強(qiáng)方法[J];西安郵電大學(xué)學(xué)報(bào);2014年02期

10 楊波;賈振紅;覃錫忠;楊杰;胡英杰;;基于Shearlet變換的遙感圖像增強(qiáng)[J];光電子.激光;2013年11期

相關(guān)博士學(xué)位論文 前1條

1 周欣;紅外圖像視覺(jué)效果增強(qiáng)技術(shù)的研究[D];天津大學(xué);2009年

相關(guān)碩士學(xué)位論文 前2條

1 游敏;基于引導(dǎo)濾波的圖像增強(qiáng)算法研究與應(yīng)用[D];華中師范大學(xué);2016年

2 高彥平;圖像增強(qiáng)方法的研究與實(shí)現(xiàn)[D];山東科技大學(xué);2005年

，

本文編號(hào)：1776265