Artistic rendering in computer graphics

Abstract

Artistic rendering is an emerging form of art technique in computer graphics which converts photographs into artistic images. There are many existing image processing software available that performs artistic rendering such as stroke-based or region-based rendering techniques, water colouring, pen and ink stroke, oil painting, cartoon rendering and etc. However, there are very few image processing software that provide different styles of edge detection for stroke distribution in artistic rendering.

In this report, a range of non-photorealistic rendering (NPR) methodologies such as edge detection is presented to introduce the various types of edge patterns useful for stroke outlining. Edge detection is a computational methodology designed to detect the boundaries of objects based on abrupt changes in the pixel intensity detected in images. Hence for images of different intensity, quality and sharpness, they will produce a unique piece of outline strokes differently. In addition, other interesting filtering effects are also presented simultaneously for successive application of effects to add in more visualization when creating artistic images.

Research studies and analysis of these techniques are first investigated by conducting experiments using two-dimensional convolution images filtering approach for pixel manipulation. Pixel intensity values change within the pixel neighbourhood as the filter applied on the image, and produces computer-generated artistic effect after filtering is done. The software is developed using SharpDevelop 3.2. These NPR techniques are presented in an educational image processing software whereby users without prior designing skills and technical knowledge can also perform image processing proficiently.

An evaluation of the resulting image produced from the edge detection and other techniques showed an amount of artistic elements added to create artistic images, with Sobel edge giving better effect than Roberts Cross, Prewitt and Laplacian edges among the edge detection techniques analysed. Finally, specimens of the resulting images rendered with different visual styles are shown and an analysis of future work for artistic rendering is described at the end of this report.