The article “Edge-raggedness evaluation using slanted-edge analysis” by Peter D. Burns presents a methodology to evaluate edge irregularity in printed images using slanted-edge analysis. This approach is based on the ISO 12233 method, commonly used to measure the spatial frequency response (SFR) in digital cameras and scanners, which focuses on the image features of slanted edges. The technique involves an analysis of the edge gradient applied to an estimated edge spread function, requiring the automatic identification of an edge feature in a digital test image.
The study particularly focuses on edge irregularity, a feature frequently considered in printed texts and images. This irregularity, also known as the tangential edge profile (TEP), is measured by evaluating the spatial deviations of edge or line boundaries. Burns describes how slanted-edge analysis can be adapted to routinely assess the quality of lines and edges.
The methodology is based on locating and analyzing the edge feature, followed by estimating the tangential edge profile (TEP). Subsequently, the RMS deviation and the edge spectrum are estimated. During edge reproduction in images, these are susceptible to various sources of degradation; minor deviations often appear blurred in certain printing systems when there are high-frequency spaced positioning errors. However, in systems such as electrophotographic and inkjet printers, under certain conditions, edges may appear rough or irregular. Engeldrum addressed this continuity from blurriness to irregularity in a model of images formed by dots, showing how dot size and overlap influence the appearance of edges.
Slanted-edge analysis was modified to evaluate edge quality, including an alternative method to find edges, avoiding derivative filtering and applying a simpler edge threshold. This method is more robust for low signal-to-noise ratio signals, minimizing bias introduced by fluctuations far from the edge.
Burns also proposes a procedure to compute a frequency-weighted irregularity measure, calculating the noise power spectrum and applying a contrast sensitivity function (CSF). This approach allows decomposing the variance of TEP fluctuations according to their visual relevance, providing a more accurate measure of visually weighted edge noise compared to standard methods that only estimate the RMS deviation without weighting.
Finally, the article concludes that the proposed slanted-edge analysis can be adopted to routinely evaluate edge quality in prints and documents, similar to its use in digital cameras and scanners. This method not only accommodates variations in print-scan alignment but also allows obtaining both the normal edge profile (to assess sharpness) and the tangential edge profile (to assess irregularity) from the same test feature and with the same software tool.
