Performance evaluation on contour extraction using Hough transform and RANSAC for multi-sensor data fusion applications in industrial food inspection

For multi-sensor data fusion applications the accurate alignment of different sensor data is essential for the proper combination of matching features. In food inspection system the boxing often is in a rectangular shape. This knowledge can be used to rectify the image data, an important step in the alignment stage. In case of low contrast between boxing and background, the detected contour may differ significantly from the actual values. In this paper the performance of the Hough transform and the RANdom SAmple Consensus (RANSAC)-algorithm are evaluated relating to the correct extraction of the boxing contour out of contour data distorted by position errors of the outer shape. The evaluation results indicate the superiority of the RANSAC algorithm with respect to scalability, robustness and execution time.