IR scene generation for the testing of image analysis algorithms

The aim of the work described in this paper, is to generate realistic infra-red (IR) images of sea clutter including sun glint, suitable for the testing of image analysis and detection algorithms. The accuracy of detection algorithms can only be determined by using a wide variety of test data representing different scene parameters such as sea state, sun angles and camera position. This type of data is extremely costly and difficult to obtain in practical trials. Synthetic generation of such scenes offers the capability to generate suitable data sets at a fraction of the cost, whilst offering precise control over the scene parameters. A model suitable for the generation of such data is developed and some results of synthetic IR scenes are presented. The model presented in this paper consists of three parts, namely: 1. The generation of the sea surface, including: animation of the wave motion. 2. The computer graphics required for the viewing of the scene. 3. The computation of the IR radiance for each element of the scene.