<title>MTF as a figure of merit in visual surface inspection</title>

Abstract

The quality of the image is crucial for the performance of a visual surface inspection system. The final image quality is affected by the properties of the illumination and imaging units. In order to evaluate the quality of these units a designer should be able to characterize their performance using quantitative figures of merit. In this paper the suitability of the modulation transfer function (MTF) for the figure of merit in visual web inspection is considered. An uncomplicated test set-up for measuring the system MTF of an imaging unit is presented and MTFs of typical web inspection imaging arrangements are introduced as a function of the f- number of the imaging lens. An appropriate set of defective samples from a paper manufacturing process is collected and imaged using the same arrangements as in the MTF measurements. Defect contrasts and signal-to-noise ratios are determined from the images and the results are compared to the corresponding MTF values. The objective of the study is to demonstrate the relationship between the MTF and the capability of inspection system to detect certain kinds of surface defects. The existence of this relationship is a necessity when MTF is considered as a figure of merit. Based on the experimental results surface defects are classified in two categories: (1) Defects whose detectivity is critically dependent on the MTF, and (2) defects whose detectivity is practically independent of the MTF. As a conclusion, examples are introduced on how the MTF measurement can be used as an imaging system evaluation and design tool.