The mandibular head has been extracted from a cine radiographic film digitized through a TV camera system, in order to trace its sagital motion during an opening and closing period of the mouth automatically. A model is given of subtraction based on the cine-film TV-input. An assumption of the additive property of radiographic images was used to compensate fluctuation of the illuminating light intensity by equalizing the average gray levels among the frames, as well as to estimate the background image by LIP, the least intensity projection. The background was estimated by peeping directly through the moving object at each pixel at an instant when the object did not contribute to the pixel intensity. This means that, for a sequence of images with both moving and unmoving objects, moving objects can be extracted by subtraction, even if it is impossible to take any image of the background separately. The developed method is called LIS, the least intensity subtraction. The mandibular head region was extracted by subsequent application of standard digital image recognition techniques including connected component extraction and erosion-dilation. The motion trajectory was obtained by tracing the top end point of that mandibular head region.
抄録全体を表示