Image analysis – Histogram processing – For setting a threshold
Reexamination Certificate
2000-02-17
2004-08-03
Bali, Vikkram (Department: 2624)
Image analysis
Histogram processing
For setting a threshold
C382S261000, C382S263000, C382S240000
Reexamination Certificate
active
06771793
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to an image processing method and apparatus, wherein processing for suppressing noise and/or processing for enhancing a desired structure pattern is performed on an original image signal representing an original image. This invention also relates to a recording medium, on which a program for causing a computer to execute the image processing method has been recorded and from which the computer is capable of reading the program.
2. Description of the Prior Art
Techniques for obtaining an image signal, which represents an image, carrying out appropriate image processing on the image signal, and then reproducing a visible image by use of the processed image signal have heretofore been known in various fields. For example, in Japanese Unexamined Patent Publication No. 55(1980)-163772, the applicant proposed a method for performing enhancement processing in a frequency domain, such as unsharp masking processing, on an image signal, such that a visible radiation image may be obtained, which has good image quality and can serve as an effective tool in, particularly, the efficient and accurate diagnosis of an illness. With the processing in the frequency domain, an unsharp 1019-1033, July 1991.
The Laplacian pyramid technique has been proposed in, for example, Japanese Unexamined Patent Publication Nos. 5(1993)-244508, 6(1994)-96200, and 6(1994)-301766. With the proposed Laplacian pyramid technique, masking processing is performed on the original image by using a mask having characteristics such that it may be approximately represented by a Gaussian function. A sub-sampling operation is then performed on the resulting image in order to thin out the number of the picture elements to one half along each of two-dimensional directions of the array of the picture elements in the image, and an unsharp image having a size of one-fourth of the size of the original image is thereby obtained. Thereafter, a picture element having a value of 0 is inserted into each of the points on the unsharp image, which were eliminated during the sampling operation, and the image size is thereby restored to the original size. Masking processing is then performed on the thus obtained image by using the aforesaid mask, and an unsharp image is thereby obtained. The thus obtained unsharp image is subtracted from the original image, and a detail image of a predetermined frequency band of the original image is thereby obtained. This processing is iterated with respect to the obtained unsharp image, and N number of unsharp images having sizes of ½
2N
of the size of the original image are thereby formed. As described above, the sampling operation is performed on the masking-processed signal is subtracted from an image signal representing an original image, the resulting difference value is multiplied by an enhancement coefficient, and the thus obtained product is added to the image signal. In this manner, predetermined frequency components in the image are enhanced.
Also, as techniques for processing an image signal, techniques referred to as multi-resolution transform techniques have been proposed. With the proposed multi-resolution transform techniques, an image is transformed into multi-resolution images, each of which is of one of a plurality of different frequency bands, and predetermined processing is performed on each of the images of the different frequency bands. Images obtained from the processing are then subjected to inverse multi-resolution transform, and a final processed image is thereby obtained. As the multi-resolution transform techniques, a wavelet transform technique, a Laplacian pyramid technique, and the like, have heretofore been known.
The wavelet transform technique has recently been developed as a frequency analysis method and has heretofore been applied to stereo pattern matching, signal compression, and the like. The wavelet transform technique is described in, for example, “Wavelets and Signal Processing,” by Olivier Rioul and Martin Vetterli, IEEE SP Magazine, pp. 14-38, October 1991; and “Zero-Crossings of a Wavelet Transform,” by Stephane Mallat, IEEE Transactions on Information Theory, Vol. 37, No. 4, pp. image, which has been obtained from the masking processing with the mask having the characteristics such that it may be approximately represented by the Gaussian function. Therefore, though the Gaussian filter is used actually, the same processed image as that obtained when a Laplacian filter is used is obtained. Also, in this manner, the images of low frequency bands, which have the sizes of ½
2N
of the size of the original image are successively obtained from the image of the original image size. Therefore, the group of the images obtained as a result of the processing is referred to as the Laplacian pyramid.
The Laplacian pyramid technique is described in detail in, for example, “Fast Filter Transforms for Image Processing” by Burt P. J., Computer Graphics and Image Processing, Vol. 16, pp. 20-51, 1981; “Fast Computation of the Difference of Low Pass Transform” by Growley J. L., Stern R. M., IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 6, No. 2, March 1984; “A Theory for Multiresolution Signal Decomposition; The Wavelet Representation” by Mallat S. G., IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol. 11, No. 7, July 1989; “Image Compression by Gabor Expansion” by Ebrahimi T., Kunt M., Optical Engineering, Vol. 30, No. 7, pp. 873-880, July 1991; and “Multiscale Image Contrast Amplification” by Pieter Vuylsteke, Emile Schoeters, SPIE, Vol. 2167, Image Processing (1994), pp. 551-560.
In a radiation image, quantum noise of radiation becomes perceptible at an image area, which corresponds to an area exposed to a low dose of radiation and which has a low image density. Therefore, various methods have been proposed, wherein multi-resolution transform is performed on an image signal, which represents a radiation image, with a technique, such as the wavelet transform, band-limited image signals falling within a plurality of different frequency bands are obtained from the multi-resolution transform, and processing for suppressing noise is performed on the band-limited image signals. The methods are disclosed in, for example, Japanese Unexamined Patent Publication Nos. 6(1994)-274615 and 9(1997)-212623.
For example, Japanese Unexamined Patent Publication No. 6(1994)-274615 discloses a method, comprising the steps of:
performing wavelet transform on an image signal by employing a second-order derivative of a smoothing function as a basic wavelet function, band-limited image signals falling within a plurality of different frequency bands being obtained from the wavelet transform,
in cases where image processing is performed on each of the band-limited image signals, detecting a point, at which a signal value of a frequency band lower by one stage than a desired frequency band is zero,
setting an enhancement coefficient such that an area in the vicinity of the detected zero point takes a value larger than the values of the other areas,
enhancing the band-limited image signal of the desired frequency band with the set enhancement coefficient, and
performing inverse wavelet transform on the thus processed band-limited image signal and the band-limited image signals, a final processed image signal being thereby obtained. Of a radiation image, major object image information is expressed in a comparatively low frequency band among the plurality of different frequency bands after the wavelet transform, and noise components are expressed in a comparatively high frequency band. Therefore, there is a strong probability that the zero point, at which the image signal of a low frequency band among the image signals falling within the plurality of different frequency bands takes a value of zero, will represent an area associated with an inflection point of the image signal representing the boundary between the major object and the other areas, i.e. an area associated with an edge area
Bali Vikkram
Fuji Photo Film Co. , Ltd.
Sughrue & Mion, PLLC
LandOfFree
Image processing method and apparatus 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 method and apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing method and apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3339911