Image-based feature detection using edge vectors

A computer-implemented method of recognizing a type of object by generating edge-based feature descriptors at a feature detection device, the method comprising:

• detecting, by the feature detection device, a plurality of edges within a digital image having medical information;
• detecting, by the feature detection device, features associated with the plurality of edges;
• selecting, by the feature detection device, at least one anchor point related to the features associated with the plurality of edges;
• generating, by the feature detection device, at least one analysis grid associated with the at least one anchor point, the at least one analysis grid including a plurality of cells;
• calculating, by the feature detection device, an anchor point normal vector comprising a normal vector of an edge associated with the at least one anchor point;
• calculating, by the feature detection device, one or more edge pixel normal vectors comprising normal vectors of the edge at one or more locations along the edge within the plurality of cells of the at least one analysis grid;
• generating, by the feature detection device, edge descriptors for the at least one analysis grid as a function of similarity measures between each of the one or more edge pixel normal vectors within a cell and the anchor point normal vector;
• and
• recognizing, by the feature detection device, a type of object according to the edge descriptors.