【摘要】：中國(guó)畫作為世界藝術(shù)寶庫(kù)中獨(dú)樹一幟的藝術(shù)創(chuàng)作形式,也是我國(guó)傳統(tǒng)文化藝術(shù)的重要組成部分。隨著中國(guó)畫逐漸步入國(guó)際拍賣市場(chǎng),近幾年出現(xiàn)了大量的贗品。所以,如何準(zhǔn)確的判斷一幅中國(guó)畫作品的真?zhèn)我殉蔀槟壳爸袊?guó)畫發(fā)展中面臨的一大難題。傳統(tǒng)的中國(guó)畫真?zhèn)舞b別主要是專家通過自身已有的經(jīng)驗(yàn),根據(jù)作品的筆法、墨法以及所處年代,對(duì)畫作進(jìn)行感性判斷,但是,這種方法極易受主觀因素干擾,并且缺乏客觀的、可量化的鑒別指標(biāo)。針對(duì)上述問題,論文使用圖像處理計(jì)算機(jī)視覺等技術(shù),并結(jié)合一些可量化的中國(guó)畫真?zhèn)舞b別指標(biāo),設(shè)計(jì)了一種基于圖像識(shí)別的中國(guó)畫真?zhèn)舞b別方法,以輔助中國(guó)畫真?zhèn)舞b別工作,提高真?zhèn)舞b別的可信度。論文主要工作分為以下兩個(gè)方面。(1)中國(guó)畫的繪畫手法獨(dú)特,風(fēng)格類型豐富,基于這一特征設(shè)計(jì)了一種感興趣區(qū)域(Region Of Interesting,簡(jiǎn)稱ROI)提取的中國(guó)畫風(fēng)格特征判別方法。新方法通過提取圖像的低階特征,分析每種特征的ROI特點(diǎn),再利用最佳權(quán)值算法,綜合提取圖像的特征ROI,最終,根據(jù)每種特征在圖像中的貢獻(xiàn)值來調(diào)整權(quán)值,以最佳權(quán)值所對(duì)應(yīng)的ROI即為特征ROI來作為圖像的風(fēng)格特征。(2)在風(fēng)格特征區(qū)域判別的基礎(chǔ)上,設(shè)計(jì)了一種中國(guó)畫真?zhèn)舞b別方法,由于馬爾科夫隨機(jī)場(chǎng)可以表示當(dāng)前值與其鄰域的關(guān)系,因此新方法建立區(qū)域級(jí)的馬爾科夫模型。其中,模型第一層利用改進(jìn)的置信傳播算法迭代遞推區(qū)域的真?zhèn)螤顟B(tài),第二層,根據(jù)區(qū)域的真?zhèn)螤顟B(tài)推出對(duì)象塊的真?zhèn)螤顟B(tài),最終,在第三層使用投票法判斷出整個(gè)作品的真?zhèn)螤顟B(tài)。最后,在Windows 7操作系統(tǒng)環(huán)境下,利用VS2012平臺(tái)下進(jìn)行了編程實(shí)現(xiàn),以驗(yàn)證方法的可行性和適用性。
[Abstract]:As a unique form of art creation in the world art treasure house, Chinese painting is also an important part of Chinese traditional culture and art. With the gradual entry of Chinese painting into the international auction market, in recent years a large number of forgeries have emerged. Therefore, how to accurately judge the true or false of a Chinese painting has become a difficult problem in the development of Chinese painting. Traditional Chinese painting identification is mainly through their own experience, according to the painting, ink and age, the painting perceptual judgment, but this method is very vulnerable to subjective factors, and lack of objective, Quantifiable identification index. In order to solve the above problems, this paper designs a Chinese painting authenticity identification method based on image recognition by using image processing and computer vision technology, and combining with some quantifiable identification indexes of Chinese painting authenticity and falsehood. In order to assist the Chinese painting true and false identification work, improve the credibility of true and false identification. The main work of this paper is divided into the following two aspects. (1) the Chinese painting has unique painting techniques and rich style types. Based on this feature, a method of distinguishing the style features of Chinese paintings extracted from (Region Of Interesting, (ROI) is designed. The new method extracts the low-order features of the image, analyzes the ROI characteristics of each feature, and then uses the optimal weight algorithm to extract the feature ROI, of the image synthetically. Finally, the weight is adjusted according to the contribution value of each feature in the image. The ROI corresponding to the optimal weight is taken as the feature ROI as the style feature of the image. (2) on the basis of distinguishing the region of the style feature, a method to distinguish the authenticity and falsehood of Chinese painting is designed. Because Markov random field can represent the relationship between the current value and its neighborhood, a new method is proposed to establish the Markov model at the regional level. The first layer of the model iterates the true and false states of the region using the improved confidence propagation algorithm, and the second layer deduces the true and false states of the object block according to the true and false states of the region, and finally, At the third level, the voting method is used to determine the true or false state of the whole work. Finally, under the environment of Windows-7 operating system, the program is implemented under the platform of VS2012 to verify the feasibility and applicability of the method.
【學(xué)位授予單位】：西安建筑科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2016
【分類號(hào)】：J212;TP391.41
，

本文編號(hào)：2452115