The evaluation of rendered color shading using edge detection method

Abstract

Color reproduction in 3D computer-generated scenes is affected by scene setting, object and camera properties, illumination and rendering technique. Regardless of controllable conditions in a computer-generated environment, the achievement of colorimetric accuracy is complex, it depends the most on hue, saturation, and lightness of the virtual object’s color and has to be considered from different perspectives. Visual perception of object color and shading is also influenced by object form, usually defined by contours. The aim of this research was to introduce edge detection method for evaluation of rendered color shading and to define how participants perceptually match color shaded renderings with predefined and proposed edge detection representations. The testing objects were color renderings of shaded objects whose colors were defined with tiff textures in the test chamber consisting of four lights, camera, and an object with shadows. In the experimental part, different edge detection methods and color channel models were used on selected colors of color renderings. It was confirmed that color, light intensity and background influence the frequency of detected edge pixels. Moreover, it was established that visual evaluation is also affected by color, background and light intensity. When the results of the frequency of detected edges and visual evaluation were compared, it was demonstrated that edge detection can be a suitable supportive method for evaluation of rendered color shading.