Self-Optimizing Image Processing Algorithm for Safety Critical Systems
Stephen Cronin, Brian K. Butka · 2018
Image processing has made many strides in the past few years in terms of accuracy and robustness. However, these advances have often times come at the cost of being understandable by humans and as a result difficult to validate the outputs. In this work we propose a genetic algorithm designed to optimize image processing while maintaining a human understandable output. The algorithm bases itself on fundamental functions of image processing and works with them to optimize the detection of objects in the frame while minimizing spurious noise. Optimization of results comes in two primary ways, through the varying the arrangement of function calls as well as through tuning the parameters used in the calls themselves. Results and conclusions drawn through the implementation of the algorithm will be discussed along with a comparison of other current state of the art algorithms.