Image analysis – Pattern recognition – Feature extraction
Reexamination Certificate
1998-05-26
2001-07-03
Boudreau, Leo (Department: 2721)
Image analysis
Pattern recognition
Feature extraction
C382S199000, C382S215000, C382S317000, C358S486000, C358S488000, C358S493000
Reexamination Certificate
active
06256411
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to an image processing device and method for recognizing and deleting extraneous objects in image data based on analysis of the image data.
BACKGROUND OF THE INVENTION AND DESCRIPTION OF THE RELATED ART
Image reading devices, often referred to as book scanners, have become commercialized. These devices generally include an image sensing system disposed above a document table so as to create an open space several tens of centimeters in height between the document table and the image sensing system. To read a bound document such as a book, magazine or the like (hereinafter referred to as “book document”), the open faced book document is placed face upward on the document table. When the start key is turned ON, the illumination lamp is activated and the sensing system starts scanning the document surface (viewing surface). As the viewing surface is scanned, the read image is sequentially output to an external device such as a printer or a storage device. When the external device is a printer, the copy image may be printed in real time as the document is scanned. Because the document is placed face upwards on the document table, the efficiency of the reading operation is improved because the pages of a book document can be turned without lifting the document from the document table.
Because the pages of an open book tend to curl or close, book documents, unlike sheet documents, must often be pressed along the edges of the viewing surface in order to maintain an open condition for scanning. In such instances, the book scanner output image may include an image of the fingers and hands of the user in addition to the document (hereinafter referred to as “extraneous object images such as fingers”). Because images of extraneous object images such as fingers are undesirable, such images must be removed before the final scanned image is stored or output. Conventionally, the extraction of extraneous object images such as fingers from a read image is accomplished by discriminating density and color pixels in the scanned image. Such methods include, for example, methods to extract an image comprised of pixels above a constant density by checking the pixel density of an object of the document margin region (white region), and methods to extract flesh colored images from an image obtained by color image sensing (See e.g., U.S. Pat. No. 5,377,019).
A problem arises in the conventional system when scanning areas that are inclined relative to the scanner. Conventionally, when an inclined area is present, such as the finger of a user, the illumination light is positively reflected by the inclined surface of the object and is thus highlighted. Because the area which is highlighted is brighter than the background of the document (i.e., has a color the color of the illumination light), the shape of the object cannot be accurately detected by either density or color. Additionally, problems can arise when attempting to extract extraneous objects based on color discrimination especially when other factors such as manicures and the like affect the color of the object.
SUMMARY OF THE INVENTION
An object of the present invention is to accurately erase the image of extraneous objects such as the fingers of a user holding a document. A further object is to increase the speed of the reading operation by reducing the time required for data processing to extract the image of objects such as fingers.
In the present invention, a dynamic contouring method for deforming an elastic imaginary line along a boundary of variable density (brightness) is used as a region separation method of a read image to extract the image of objects such as fingers. By setting suitable elasticity, the imaginary line converges to a curve which smoothly circumscribes the image of an object such as a finger. As a result, the image of an object such as a finger can be erased by obtaining an accurate contour of the actual shape of an object and the pixel value on the interior side of the obtained contour can be substituted with a value equivalent to the background even in the case when positive reflection of illumination light causes an abnormal density distribution.
The initial value of the imaginary line is initially set for efficient extraction. Since objects such as fingers holding a document typically appear in the margins of the document, a straight line or curved line arranged near the edges of the document is preferably set as the initial value.
According to one embodiment of the present invention, an image reading device is provided with an object detection means for detecting an object region of a finger or the like in a document surface based on the read image of the document, and an image processing means is provided for erasing part of an image corresponding to the object region. The object detection means extracts the object region by use of a dynamic contouring method where a specific initial contour corresponding to the image is set such that the initial contour is deformed along the boundary of variable density in the read image.
In another aspect of the present invention, the image reading device is provided with a means for detecting the edge of the document, and the object detection means sets the initial contour to a straight line with fixed endpoints having a deformation tendency extending to the interior side along the document edge and having a position on the exterior side of the detected edge.
In another aspect of the present invention, the image reading device is provided with a means for detecting the edge of the document and a means for detecting a specific density area which straddles the detected edge in the read image. Here, the object detection means sets as an initial contour a curve through the first and second intersection points of intersection of a detected edge and the contour of a detected specific density area.
These and other objects, advantages and features of the invention will become apparent from the following description thereof taken in conjunction with the accompanying drawings which illustrate specific embodiments of the invention.
REFERENCES:
patent: 4980781 (1990-12-01), Yamamoto et al.
patent: 5377019 (1994-12-01), Okisu et al.
patent: 5487116 (1996-01-01), Nakano et al.
patent: 5528387 (1996-06-01), Kelly et al.
patent: 5706419 (1998-01-01), Matsugu et al.
patent: 5742279 (1998-04-01), Yamamoto et al.
patent: 5764383 (1998-06-01), Saund et al.
patent: 5832116 (1998-11-01), Rezzouk
patent: 5862245 (1999-01-01), Renouard et al.
patent: 06251148 (1997-09-01), None
Liu, et al. “optimal contour approximation by deformable piecewise cubic splines”, IEEE, pp. 638-643, 1991.
Boudreau Leo
Mariam Daniel G.
Minolta Co. , Ltd.
Sidley & Austin
LandOfFree
Image processing device and method for detecting objects in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image processing device and method for detecting objects in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing device and method for detecting objects in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2435157