Analysis of a captured image to determine a test outcome

摘要

Determining a “test outcome value” by: acquiring an image of a “test structure”; determining a mapping (registration) that transforms a “shape template” for the test structure to align it with the captured image; determining whether the mapping satisfies a first matching condition and if so whether it satisfies a second matching condition to “verify” the image, wherein both matching conditions are based on displacements of edges identified in the image relative to the shape template; identifying a “target measurement region” in the verified image and analysing it to determine a test outcome. Mapping may be iterative, employ machine learning, and use optical flow tracking for time series of images. Matching may be based on mean squares of distances between edges or points of acquired images and reference images. May be used in medical, diagnostic, biological, microbiological, biochemical or chemical assays based on a colour change over time. Reaction colours may be compared against a reference chart (eg. detecting glucose, ketones). May test for skin colour variations associated with skin cancer or raised bilirubin levels. Test equipment may be sensitive to atmosphere or temperature. Photographs may be captured using a mobile device such as a phone.