| 摘要 |
PROBLEM TO BE SOLVED: To extract an object at a highs peed and with high accuracy. SOLUTION: In this image extraction method, an initial kernel setting device 20 sets a point or an area that is designated by an instruction selection device 18 as an area growing kernel. A feature vector extraction device 22 extracts the color component value of every pixel of an input image, the edge intensity value concerning density (lightness), the contrast component or texture feature vector concerning lightness, etc. A key color extraction device 24 extracts the color component feature vector of the key color from a point set on the initial kernel. An area division device 26 labels the pixels having small differences between one of key colors and the color components as the continuous areas and generates the image division data to specify every area. An area growth degeneration device 28 merges or excludes each of points of areas near the boundary set to a set initial kernel area or an area whose shape is changed in or from an attentional area according to the image feature vector or the image division data. A shape display data generation device 30 generates the display data showing the shape of an area which is obtained in an area growing process (or from the area growing result). Then an image based on the display data is shown on the input image. |
