There have been efforts to detect the collimation field of medical digital radiographs (x-ray images), which is unnecessary in diagnosis. The method used in past studies is to detect the boundary of the collimation field as a straight line, using the edge detection on the two-dimensional image. The boundary, however, is not an ideal step-edge, as is seen from the principle of image generation, but is a ramp edge containing a blur (half-shadow). This paper extends the radiograph to another dimension of brightness, to be considered as three-dimensional data. A method is proposed in which the Hough transform is used to detect the, boundary of the collimation field as a plane. In order to investigate the validity of the model and the effectiveness of the proposed method, a detection experiment is performed for the collimation field. The collimation field is correctly detected at a rate of 96.4% (324 cases) in 336 images. It is also shown that when the image is compressed by setting the brightness of the detected collimation field as 0, the amount of image data is smaller than the amount of image data obtained by compressing the original image. © 2006 Wiley Periodicals, Inc.