| 摘要 |
A method of extracting a texture region from an objective picture is disclosed. The method comprises the steps of determining a direction of a line image represented by pixels lying adjacent each of all pixels forming the objective picture by using a spatial filter to define the determined direction as a direction of each of all the pixels, determining the number of the pixels showing the same direction for each of the preselected directions, and extracting from the objective picture as the texture region an area where the number of the pixels showing the same direction for at least one of the preselected directions is greater than a given value and a dispersion value of the pixels showing the same direction is greater than a given dispersion value. A second method is disclosed which detects the texture region based on the fact that pixels assuming a particular direction usually exist on a high-frequency portion such as the texture region or the contour of a non-texture region. A third method is disclosed which detects as the texture region an area where end points of line images exist at a density greater than a given value. <IMAGE> |
