Image Correspondence Figure of Merit (FOM)

Aret Associates developed an algorithm that takes as input 2+ electro-optic (EO) images of a target area, and outputs a Figure of Merit describing the performance of an image recognition algorithm which matches input infrared (IR) images with reference imagery. This algorithm is designed to determine the angle of approach enabling the best image recognition. Aret implemented a prototype version of the algorithm, which detects line features, allows the operator to associate line features across images, then finds the three-dimensional coordinates of the features. The line feature set is characterized in terms of descriptive attributes; this approach was demonstrated on several datasets.Aret will deliver a GUI-driven beta version of the FOM software by the end of Phase II. In addition to the GUI, Aret will further develop the line association algorithm; we will create a cross-modal association metric to quantify the match between two line feature sets; we will refine the feature characterization algorithm; finally, we will design an initial Figure of Merit formula that outputs a value based on the feature characteristics. In the Phase II Option, Aret will validate the feature characterization and Figure of Merit formula using a large cross-modal imagery dataset.