해양 쓰레기 모니터링 시스템 구축을 위한 칼라 영상 클러스트링 기법의 개발

Abstract

A color image clustering algorithm is proposed for marine debris monitoring system. The color image clustering technique is composed by mahalanobis distance to detect the object region, where marine debris is accumulated from the obtained input image by using a pilotless aircraft about the monitoring area of a study region. Two color models the forest area and the sand area are selected for mahalanobis distance calculation, and each model is compared with similar color regions from the input image, respectively. The binarization called Otsu method is applied on two sets of mahalanobis distance about the forest area and the sand area. The object region with a certain width is detected from the boundary that is extracted from common information between the forest area and the sand area. The effectiveness of the algorithm is examined using series of input images from the actual monitoring area. As the result, it is confirmed that the algorithm is useful technique to construct the marine debris monitoring system.