航海自然画像におけるフラクタル次元値の推定と領域分割への応用

概要

論文の詳細を見る
Fractal dimension (FD) values can quantify the complexity about the shape of surfaces and curves. In this paper, we propose a method of region segmentation based on the FD for natural navigation images, which are the digital images of natural scenes viewed around at a ship's bridge in passing a narrow channel. The region segmentation means that the image is divided into physically meaningful objects, i.e., sea region, land region and sky region and others. In this method, the regions on the images are treated as the texture and the regions are segmented by the difference of the texture complexity. Then, the FD is used as the scale for measuring the texture complexity. The threshold values used in the segmentation are estimated from the averages between the FD values calculated on each block that are small areas on the region.
社団法人日本航海学会の論文
1994-03-25