素描筆觸資料庫及筆觸對應演算法

Abstract

在鉛筆素描之模擬範疇中，本論文旨在探討如何產生筆觸資料庫以及如何根據圖片區域特性，從筆觸資料庫中調用適合的筆觸在區域上。做法主要有三個步驟：1.運用分層演算法將圖片的灰階度加以分層。2.計算不同層區域的特性。3.依據區域特性從筆觸資料庫調用合適的筆觸。 本論文提出的筆觸資料庫乃依照不同的筆觸參數創建的筆觸資料庫，使研究員可以專注在研究區域特性，以及根據區域特性調用資料庫中不同的筆觸。

Parameters controlling stroke presentation in pencil sketch simulations include at least orientation, length, width, inter-stroke distance, stroke-pixel density, shade. While sketching with pencils, dealing large areas in the drawing with wide strokes in a shade of deem contrast is a common practice. In this thesis, such artistic presentation in large area sketching is simulated by applying strokes of relatively large width with proper distance in between, in addition to the tuning of pixel density along the stroke trajectories and minute contrast adjustment at stroke-pixel. Key issues addressed in this thesis include: [1] image segmentation discouraging unnecessary dissection of large homogeneous area is devised; [2] the visual sensation caused by the cross interactions among segment area sizes, stroke widths, inter-stroke distances, stroke-pixel densities, and contrast variation around a stroke-pixel is thoroughly examined in a systematic way; [3] a database of sketch strokes is thus established; [4] an algorithm for acquiring a proper stroke (orientation, width, stroke distance, stroke-pixel density, stroke-pixel contrast) In the data base when drawing any segment area in the sketch is developed;

A pilot system for pencil sketches simulation is developed and the outcomes of most bench tests are found to be visually satisfactory.