Using morphological transformations to recognize ships and vessels

The article proposes algorithms for improving image segmentation with a focus on morphological transformations to solve problems of recognition of marine vessels. Particular attention is paid to the binarization of images and its role in the identification of marine vessels. The developed algorithms are investigated and evaluated using the example of segmentation of transport plots. The results of contour selection based on such operators as Sobol, Prewitt and Canny are presented. The proposed algorithms were found to provide for higher accuracy of recognition of marine vessels by an average of 15 % compared to the known recognition methods based on the Viola and Jones method on Haar Wavelets and by 12 % compared to the recognition method by reference points.

Authors: S. Sh. Fahmi , N. A. Ponomarev, M. T. M. Ho

Direction: Informatics, Computer Technologies And Control

Keywords: segmentation, morphological transformations, binarization, erosion, dilation, marine vessel recognition

---

View full article