Localization of moving edge with sub-pixel accuracy in 1-D images and its FPGA implementation

Contour detection is an algorithm often utilized in picture processing. Sometimes it is useful to localize edges with sub-pixel accuracy. Many methods have been developed for edge detection with sub-pixel accuracy. The question is, how the accuracies of these methods change if the scanned object is moving during the exposure time. In this paper, the impact of object movement on edge detection accuracy is examined. To simulate the moving edge, the accumulative function is defined and then used in three edge detection algorithms with sub-pixel accuracy in 1-D images: algorithm based on approximation of image function with function erf (AEF), method based on statistical moments (GLM) and technique using spatial moments of the 1-D image (SM). Results of simulations with noisy images are presented, the upper and lower 5% quantiles are chosen as accuracy criterion. Gray level moment edge detector (GLM) is used for FPGA implementation, because of its accuracy and simplicity.