A method and apparatus for extracting objects from an input image allows a user to select seed coordinates of the desired object. An edge detection component detects an edge of the object by comparing the intensity of a pixel to its surrounding pixels. A seeded region grower creates a segmented object image by converting the input image and the edge-detected image into gray scale data and further binarizing the edge-detected gray scale data. The seeded region grower creates a base region beginning with the seed coordinates and grows this region until it includes an outline of the object. A containment component then maps the segmented object image in four separate directions to tag pixels located in the desired object. An output component creates the output image containing the object by selecting the tagged pixels from the original input image.
