Image analysis – Image enhancement or restoration – Image filter
Reexamination Certificate
1998-12-29
2002-10-15
Lee, Thomas D. (Department: 2724)
Image analysis
Image enhancement or restoration
Image filter
C382S266000
Reexamination Certificate
active
06466700
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to an image processing method for reducing the noise while preserving the edges of objects reproduced in an intensity image, which method includes a step for the acquisition of the intensity image in the form of a multidimensional matrix of points The invention also relates to a medical imaging apparatus which includes a system for carrying out this method.
Points in a multidimensional matrix are to be understood to mean pixels in the case of a two-dimensional image and voxels in the case of a three-dimensional image.
The invention can be used notably in medical X-ray systems producing multidimensional images.
BACKGROUND OF THE INVENTION
A method of reducing the noise in a noisy digital image is already known from the publication “Image Processing: Flows under Min/Max Curvature and Mean Curvature” by R. Malladi et alii, in GRAPHIC MODELS AND IMAGE PROCESSING, Vol.58, No. 2, March, pp.127-141, 1996, ARTICLE No. 0011.
This document describes a class of algorithms which are suitable for reducing the noise in an image and for enhancing the contrast thereof. The method can be applied to noisy black and white images or color images. The method comprises two principal steps. First of all, a formulation of a set of grey levels or intensity levels is used to generate curves linking the pixels of the same intensity in the image; these curves are called isophotic lines. Secondly, means are used to smooth the irregularities of the curvature of the isophotic lines, thus reducing the noise and enhancing the images. The operation for smoothing the isophotic lines is carried out by reducing minima and maxima of curvature of the isophotic lines; this is called curvature commutation. The degrees of reduction of curvature are selected in a window so as to diminish the irregularities of the pattern of isophotic line within boundaries. As a result, the noise is reduced in this commutation window. This method is recommended according to the cited document first of all because it includes only one enhancement parameter which is automatically chosen in the majority of cases; secondly, it is recommended because it automatically stops the smoothing of a point in dependence on the dimensions of the commutation window.
Actually, this method is based on iterative steps An iterative step is started anew in a loop as many times as is necessary to obtain correct smoothing of isophotic lines. For a complex image, i.e. a non-geometrical image, and a three-dimensional image, that is to say an image consisting of, for example 128
3
voxels, however, tests have shown that it is necessary to perform from 300 to 500 iterations which, according to the present state of the art, leads to calculation times of the order of 10 hours per image.
SUMMARY OF THE INVENTION
It is an object of the invention to provide a method for eliminating the noise, i.e. for smoothing a multidimensional complex image, which requires a much shorter calculation time, i.e. a method which requires only a small number of iterations.
This object is achieved by means of an image processing method for preserving the edges of objects reproduced in an intensity image and for reducing the noise by acquisition of the intensity image (2-D, 3-D) in the form of a multidimensional matrix of points, determination of the intensity gradient at each point, defined by its direction (&thgr;
p
) and its modulus (∥G
p
∥) , and passage at each point of a recursive spatial filter having an anisotropic kernel and providing a degree of smoothing which is greater in the direction perpendicular to the direction to the gradient than in the direction of the intensity gradient;
An advantage of the method according to the invention resides in the fact that it is recursive, so that it can be carried out very economically in respect of calculations because each application of the filter to a point utilizes calculation elements which have already been performed. Thus, it can be executed -much faster than an iterative method. For example, using contemporary techniques the present method requires less than one minute for processing a volume of 128
3
voxels. Therefore, it is much easier to carry out and approximately 1000 times faster than the known method.
Another advantage of the present method resides in the fact that it is adaptive, meaning that the filtering is automatically adapted to the type of structure encountered in each point. Thus, this method applies different filtering to the points of the background and the points of the object edges. The method thus enables smoothing of the image while preserving neat, non-shifted, non-distorted object edges wherefrom the noise has also been removed. This method is exact and powerful.
It is another advantage of the present method that it concerns a spatial filter. Thus, it does not require any data other than that of the single image processed; moreover, it is robust. Because it is very fast, it can also be used for the processing of a sequence of images.
Another advantage of the present method consists in the fact that it is fully automatic.
This method is carried out in an apparatus including a system for the acquisition of an intensity image in the form of a multidimensional matrix of points, and an image processing system which has access to the image data and to an image display and which includes a microprocessor for carrying out the image processing methods of the invention.
REFERENCES:
patent: 5673207 (1997-09-01), Nomura
patent: 6072496 (2000-06-01), Guenter et al.
patent: p 927 953 (1999-07-01), None
Brinich Stephen
Koninklijke Philips Electronics , N.V.
Lee Thomas D.
Vodopia John
LandOfFree
Method of processing a noisy multidimensional image and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of processing a noisy multidimensional image and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of processing a noisy multidimensional image and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2947356