HSV分析による航海環境時系列画像からの船舶抽出

概要

論文の詳細を見る
This paper proposes detection method of ships using navigational image sequence analyzed by HSV transform. The region constructed with same color can be detected for specifying a Hue range. But it is difficult to decide a color of a target ship. So we proposed detection methods of ships by distribution of Hue variance. We verified the advantages of the proposed methods by processing images captured at the Tokyo port and Tokyo Bay. As a result, the operator method was judged to be the most excellent. The operator method divides the image into a small rectangular area, and to calculate the variance of Hue in the area. The areas that include ships have many colors, so the variance will be high. We scanned the image with operator and check the variance, because the area with high variance includes ships. It turned out that it was not possible to detect well in the Rain and Fog conditions. As for these conditions, we can solve by taking a picture by the infrared camera simultaneously and using value parameter.
社団法人日本航海学会の論文
2007-03-25