Image generation techniques for out-door scene simulation

摘要

Scene simulation has an important role in numerous applications such as the design of machine vision systems, the design of virtual environments and the test and validation of image processing algorithms. Many approaches have been proposed,however the majority of them are intended for indoor scene simulation. The simulation of outdoor scene is more difficult problem due to the influence of weather and temperature conditions. For the simulation of outdoor scenes, a gray level imagegeneration approach based on mean and variance prediction is first presented which simulates the scene under a hypothetical set of weather conditions by using a transformation of the same outdoor scene under a different set of known weather conditions.Secondly, an image synthesis approach via linear interpolation of standard sample images is derived from Kajiya's rendering equation theory. Finally a simulation approach for outdoor scene in rain and fog is given based on the relationship between thepoint spread function (PSF) and modulation transfer function (MTF) and the quality of imaging. A number of experiments illustrate that the similarity of our simulated imagery to the captured images of an outdoor scene under variable weather conditions.