Algorithm for point-to-point correlation of geometrically nearly similar microscopic objects

An algorithm is presented that compares two quasi similar images by correlating selected points on them--assuming their coordinates are available. The procedure involves translational, magnificational and rotational operations to find corresponding point pairs on the pictures. The algorithm automatically compensates for slight dissimilarities between images and constructs a reference point database for correlation during the evaluation process. Establishment of the reference point networks on the images prior to the examination is avoided.