SEARCH FOR SMALL OBJECTS ON A BACKGROUND OF A HETEROGENEOUS RANDOM TEXTURE
A. V. Likhachev
Institute of Automation and Electrometry, Siberian Branch, Russian Academy of Sciences, Novosibirsk, Russia
Keywords: search for objects in the image, object with a size of ten pixels, brightness histogram, search for the maximum product of probabilities
Abstract
An algorithm aimed at searching for small-sized (about ten pixels) objects with constant brightness on a non-uniform random background has been developed. The algorithm is based on the procedure for finding the conditional maximum of the product of probabilities for an object and a local background to have certain brightness values. For the background, these probabilities are estimated from a histogram constructed for an image segment. The algorithm is implemented for the case when all the brightness values of an object in the acceptable range are equally probable. The computational experiment is performed on twenty photographs of clouds with a size of 1200 ⨯ 1200 pixels. Ten circles with a radius of two pixels are placed on each photo. Of the two hundred objects set in this way, 169 are detected by the proposed algorithm, while the number of errors of the second kind is 280.
|
