ベクトル場によるカラー画像の特徴抽出
1987 年 11 巻 20 号 p. 37-42
詳細
1987 年 11 巻 20 号 p. 37-42
This papaer describes a new feature extraction method from color images. The chrominance signals are considered to be a vector field of two dimension. In the vector field theory, any vector can be divided into two components, lamellar and vortex. Lamellar component describes potential field and vortex describes the rotation of the field. For the color image, lamellar component represents the global structure of the image and vortex component represents the boundary information. Using these properties, we propose the new coding scheme which is similar to plateau coding.
