使用自動挑選區塊及色彩和諧化方法之藝術家畫風轉換研究

彭郁銘
Painting style transformation aims at producing a painting of one artist’s style from picture or other paintings. In the previous works of example-based painting style transfer, users have to select patches which describe painting style best from example patches manually. It is time-consuming and subjective for a user to select proper patches from a large number of patches.

This paper developed a patch-based approach for rendering example-based images without requiring user intervention to find appropriate patches in the synthesis process. We use mean shift segmentation and texture re-synthesis methods to construct the artistic database which enables users to synthesize images according to selected painting style. Moreover, seven important painting features are proposed in finding adequate correspondences between source image and database. The synthesized output images are generated by patch-based sampling method after the correspondence has been found. We also introduce a color harmonization method to change the original result according to different color harmony scheme. Experimental results show the feasibility of the proposed approach by demonstrating the synthesis results from different types of source images to painting style of Vincent van Gogh.

畫風轉換是把一張影像轉換成某個畫家風格的研究。在前人的研究中，如果想要進行畫風轉換的話，使用者必須從畫家的作品中選擇一到數塊可以代表畫家風格的區塊。想要從很多張畫家的作品中找到適合的區塊，是相當費時而且主觀的。為了解決這個問題，我們提出了一個自動從資料庫中，選擇適合區塊的方法。首先，我們將畫家的畫作們進行平均值移動演算法的影像切割，切割出來的區域分別再使用材質重新合成的方法，就可以得到一塊塊畫家風格的區塊，所有的區塊將會儲存在畫家的資料庫中。接下來，我們使用了七種有關材質特徵的計算，讓系統能從畫家資料庫中自動的找出適合目標影像的畫家風格區塊。最後透過基於區塊樣本的材質合成技術來得到最後的結果。除此之外，我們引入了色彩和諧化的理論，讓原本產生的畫家風格影像，根據不同的和諧方案，產生不同感覺的色彩和諧結果。實驗的結果顯示我們提出的系統的確可以幫助使用者從數百張中選出適合目標影像的畫風區塊，並且產生具有畫家風格的影像。