Abstract
Surface texture, like colour, is a critical aspect of product appearance. Yet, unlike colour, it cannot be easily captured, measured, or visualised using commonly available tools. While a buyer can confidently measure and electronically communicate the colour of a new design to an offshore garment manufacturer – the same cannot be said of texture. The first part of this presentation will describe how consumer level hardware can be used to capture and convincingly display surface textures in real time over the web. In the second part I will present a measurement model for perceived roughness that provides a first step towards developing a measurement system for surface texture.

Abstract
I will present a simple first order model of how variation in illumination affects the output of Filter Response Filters (FRF). FRF are of interest because: (a) they are commonly used as texture features in automated texture classification systems, and (b) they are typically proposed as the “back pocket model” of the first stage of the human visual system. I’ll show how naïve classifiers built using these simple features can fail, and how the model can be used to produce a classifier that is robust to illumination variation. What this will show is that single still images are not often not sufficient for the purposes of surface classification – either for human or automated systems. I’ll conclude by describing some of our recent research that is investigating our perceptions of surface texture.

Abstract
The image of a camouflaged animal on a background surface provides the input to a predator’s visual system and is the starting point for neural processing that will determine whether the prey is detected or not. The spatial pattern of luminance in the image arises from variation in both the reflectance and the surface relief of the animal and its background. It is also a function of the illumination of the scene, and will vary with the azimuth and elevation of the light source, and the amount of ambient light present. The effects of these variables of illumination on the image of a camouflaged animal and its background will be important determinants of its visibility to a predator. Graphics techniques such as rendering of surface height maps and bidirectional texture functions provide methods for exploring these effects in either real or synthetic textured target and background surfaces. Some examples are described, and their implications for the effectiveness of different types of camouflage under varying natural illumination conditions are discussed.

Abstract
Testing saliency models for the control of visual attention requires stimuli which, unlike photographs of natural scenes, contain no semantic information. We present a novel method for constructing such stimuli, by rendering height maps of 1/fâ noise surfaces. The resulting images strongly resemble surfaces of natural materials. Observers’ search times and fixation paths were recorded while searching for targets in these images, and were compared with the outputs of Itti and Koch’s (2000) visual saliency model. The results demonstrate that the model closely matches human performance in an ecologically valid active viewing task.