Method for segmenting aerial images based region and Computer readable storage medium for storing program code executing the same

摘要

The present invention relates to a method for segmenting an aerial image for an unmanned aircraft image and, more specifically, to a method for segmenting an aerial image based on a region which classifies a relevant terrain into a class to which the relevant terrain belongs, and improves the accuracy of the classification using context information from intelligent image processing field, such as monitoring, reconnaissance, etc. for general outdoor images and aerial simulation images, and terrain analysis field by combining an existing technology which is generally used in image processing and a newly designed technology, and developing an image segmentation technology optimized for aerial images. The method for segmenting an aerial image comprises: an initiation step of applying a filter bank which is a set of filters to pixels of an aerial image, and extracting filter responses; a response obtaining step of obtaining responses by segmenting the filter responses into a narrow region and a wide region and applying a segmenting technology; a pre texton generation step of extracting a texton by textonizing the aerial image for each pixel and generating clusters, generating a texton map by mapping each of the filter responses onto a cluster center point, and generating a response level using the texton map; a boosting step of extracting final responses of each pixel in the aerial image by minimizing errors about the response level and giving weightings; and a category determination step of classifying a relevant pixel among the pixels in the aerial image into categories according to the highest rate among the extracted final responses.