相片之自動化鉛筆式素描顯像

Abstract

本篇論文的研究目的在於自動化產生鉛筆素描效果圖，讓即使不擅長美工的使用者，也能在短時間內得到令人滿意的結果。本篇論文所產生的素描風格偏向輪廓與結構特徵的加強，所提出之自動化產生鉛筆式素描效果圖的方法分為三個重點。第一個是針對亮度較暗的區域作著色式素描。第二個是將輪廓偵測的結果以較暗的筆觸突顯。第三個重點是對那些相對於周圍較暗的結構特徵區域作著色式素描。三個重點素描結果經權重加成後，再結合紙張材質效果，構成鉛筆式素描圖輸出。

The thesis presents an automated system that is capable of generating satisfactory pencil drawings without intervention from users of sketching skills. The system consists of three processing components, respectively taking care of the parallel/cross hatching of low-intensity areas in the image, line drawing of subject silhouettes and the drawing of local contrast associated with the subjects that reflects structural features in general. The outcomes of the three components are then fused by appropriate weighting, to which the effect of paper texture is applied for a final pencil drawing with the silhouettes and structural features of the subjects being emphasized.