Interactive Rapid Prototyping based on Computer Graphics for the Image Acquisition in Automated Visual Inspection

点击翻译按钮获取中文摘要

In this project, we will research and develop basic concepts and methods to use synthetically generated images for reducing the high effort for design, development, implementation, and testing of systems for automated visual inspection (AVI) and to increase their performance and robustness. Our new approach comprises a fundamental and systematic investigation of the use of synthetic images, created by means of computer graphics (CG), for these purposes. To this end, our goal is to create a solid scientific foundation for simulation approaches in general, to be achieved by developing and defining appropriate methods and criteria. These results are expected to find broad applications in numerous tasks of image processing systems, and also more generally in the field of sensor simulation. In particular, they will facilitate and improve the design and evaluation of high-performance and robust AVI systems for applications in safety critical or quality sensitive areas.In contrast to previous approaches, we will utilize information which is available from the image generation process for a guided, and thus significantly improved and accelerated, optimization of AVI systems. As an example, the visibility between objects and illumination as well as the entire flow of light within the setup are known during the image synthesis in CG and can be used to find the optimal image acquisition setup of the AVI system.Another difference to previous works is the introduction of an interactive and (partially) automated optimization process for the image acquisition components extrinsic and intrinsic parameters as well as the optimal number of images for image sequences. Within this process, first the visual inspection method which will be used is selected by an AVI-expert. In order to optimally determine the numerous parameters of the acquisition setup, an automated parameter optimization based on the information obtained from the image synthesis is performed; the inspection method is only modified by the expert, when she/he decides to do so because of unsatisfactory results. This task sharing combines the competence of the expert and the advantages of an automated optimization based on CG in a productive way.In this proposed joint project, the cooperation of three institutes, each of them providing relevant expertise in the respective research area, makes it possible to explore this interdisciplinary topic with a thoroughness which has not been reached by previous approaches. In consequence, we expect results which are also applicable for a wide range of applications.