本文選題：圖像修描 + 水平集　； 參考：《西安理工大學(xué)》2006年碩士論文

【摘要】： 為了恢復(fù)受損的圖像，須對其受損區(qū)域進行修描，以滿足人們對高質(zhì)量印刷復(fù)制技術(shù)的需求，本課題針對該類圖像中的結(jié)構(gòu)型破損圖像設(shè)計了一種基于水平集方法的圖像修描技術(shù)。該技術(shù)可以應(yīng)用到考古、藝術(shù)、軍事、廣告等領(lǐng)域，它既提高了圖像修描的質(zhì)量，同時也改善了圖像的視覺效果，尤其對文物保護和歷史資料復(fù)原具有十分重要的意義。 本文總結(jié)了圖像處理理論在圖像修描中的應(yīng)用，通過分析人類視覺系統(tǒng)中的心腦認(rèn)知機制，總結(jié)出其對圖像修描的一些啟發(fā)性原則，將一般的圖像恢復(fù)理論應(yīng)用到圖像修復(fù)領(lǐng)域，以最佳猜測原理和貝葉斯結(jié)構(gòu)體系為理論框架，，建立了基于偏微分方程和水平集方法的圖像修描模型。本文將修描過程視為修描邊界曲線的向內(nèi)演化過程，在修描邊界向內(nèi)收縮的過程中完成對待修描區(qū)域的恢復(fù)。同時，我們選取待修描區(qū)域邊界外部的一系列點來修補圖像，這樣就避免了待修描區(qū)域內(nèi)的點對修描結(jié)果的干擾。修描公式中選取了合適的權(quán)重，保證了該修描技術(shù)與手工修復(fù)圖畫的過程很類似，從而可以達到較好的修描結(jié)果。 試驗結(jié)果表明：與現(xiàn)在流行的一些修描算法相比，本文使用的快速步進算法應(yīng)用比較簡單，并且修描效果較好，運行時間很短，特別是在其實用性方面有了很大的提高。在文章的最后，作者總結(jié)了本文設(shè)計的圖像修描技術(shù)存在的問題和不足，并對以后的研究方向做出了說明。
[Abstract]:In order to restore the damaged image, it is necessary to repair the damaged area in order to meet the needs of high quality printing and reproduction technology. In this paper, an image repair technique based on the level set method is designed for the structural damaged images of this kind of images. This technique can be applied to archaeology, art, military affairs, advertisement and so on. It not only improves the quality of image repair and drawing, but also improves the visual effect of image, especially for the protection of cultural relics and the restoration of historical data. In this paper, the application of image processing theory in image redrawing is summarized. By analyzing the cognitive mechanism of heart and brain in human visual system, some enlightening principles of image correction are summarized. The general image restoration theory is applied to the field of image restoration. Based on the theory framework of optimal guess principle and Bayesian structure system, an image correction model based on partial differential equation and level set method is established. In this paper, the redrawing process is regarded as the inward evolution of the boundary curve of the trimming boundary, and the restoration of the repair area is completed in the process of the redrawing boundary shrinking inward. At the same time, we choose a series of points outside the boundary of the area to repair the image, thus avoiding the interference of the points in the area. The appropriate weight is chosen in the formula to ensure that the process of repairing the drawing is very similar to that of the manual repair of the drawing, so that a better result can be achieved. The experimental results show that compared with some popular correction algorithms, the fast step algorithm used in this paper is relatively simple in application, and its effect is better, and the running time is very short, especially in its practicability. At the end of the paper, the author summarizes the problems and shortcomings of the image repair technique designed in this paper, and explains the research direction in the future.
【學(xué)位授予單位】：西安理工大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2006
【分類號】：TP391.41

【引證文獻】

相關(guān)碩士學(xué)位論文 前1條

1 何靜;非紋理圖像修復(fù)方法研究[D];西安理工大學(xué);2009年

本文編號：1979751