Step Edge Location with Subpixel Accuracy in Gray Level Images from Binary Scenes O. Reinoso, R. Aracil, L.M. Jiménez, C. Fernández, J.M. Sebastián, F. Torres Abstract. This paper describes a local method to locate step edges with subpixelaccuracy in low-noise images. An algorithm that allows to get higher resolutions in step edges in gray level images is proposed. The proposed local method takes into account the intensity values (in gray level images) in 5x5 neighborhoods of each pixel. A quadratic function over the middle pixel of these neighborhoods models each step edge. Keywords: Sept Edge Detection, Subpixel Accuracy, Image Reconstruction.