This paper presents a mathematical model and a geometric method of image-based object classification intended to improve the accuracy of object classification by filtering out noise objects.

Keywords: image processing, object recognition, pattern recognition, object identification, classification

The article presents the implementation results of the methods of linear contrasting, image histogram equalization, adaptive histogram equalization (AHE) and contrast limited adaptive histogram equalization (CLAHE) as test software for thermal image processing. It also proposes some ways of enhancement of the CLAHE method that eliminate the necessity to manually select the parameters to be changed.

Keywords: image processing, computer processing, histogram processing, histogram equalization