Image analysis – Image enhancement or restoration – Image filter
Reexamination Certificate
1998-12-22
2002-03-19
Rogers, Scott (Department: 2624)
Image analysis
Image enhancement or restoration
Image filter
C382S262000, C382S264000, C348S620000, C348S607000
Reexamination Certificate
active
06360025
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to an image processing method for reducing the noise in an image, including the determination of three temporal intensities relating to a current pixel in the same location in three successive images in a sequence, the image to be processed being the central image of the sequence. The invention also relates to a medical apparatus which includes such a system.
The method is applied so as to reduce the noise in a sequence of images while preserving the small details of objects in motion. Small details are to be understood to mean herein objects of from 1 to some pixels, for example from 1 to 10 pixels. The invention can be used particularly for the processing of video images, notably medical images.
2. Description of Related Art
An image processing method for reducing the noise while preserving small details is already known from the publication “A new class of Detail-Preserving Filters for Image Processing” by ARI NIEMINEN, PEKKA HEINONEN, and YRJO NEUVO in IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL.PAMI-9, N°1, JANUARY 1987, pages 74-90. The cited document describes, in a general manner, two-dimensional spatial filters which include a median which is applied to a structure with a finite impulse response called FMH-2D (the letter F stands for FIR or Finite Impulse Response; M stands for Median, H stands for Hybrid; 2D stands for two-dimensional). Referring notably to FIG.
4
and section II-C of the cited document, a Finite Impulse Response (FIR) filter of the type FMH-2D comprises four small linear masks, each of which consists of two pixels which form four branches which are arranged at an angle of 90° around a current pixel in the plane of the image. The filtering includes a step for averaging the intensity of the two pixels in the small masks, and a step for classifying the mean intensities obtained in the four branches and the intensity of the central current pixel by a median which performs the selection of one of these intensities. Another filter, called a FMH-1D filter and described with reference to FIG.
2
and in section II-B, involves one or several median levels. The FMH-1D filter involves two elementary masks, arranged in a single direction in the plane, and the realization of a median over the mean intensities in the masks and the intensity of the central current pixel. A problem is encountered in that these exclusively spatial filters produce exclusively an average noise reduction, so that they still leave behind noise patches and noise patterns.
SUMMARY OF THE INVENTION
It is an object of the invention to provide an image processing method which includes a filtering operation which provides an improved noise reduction without noise patches and noise patterns in the movements and which notably does not erase the small details of moving objects in a sequence of images.
This object is achieved by means of an image processing method including determination of three temporal intensities relating to a current pixel in the same location (x,y) in three successive images ( J
t−1
, J
t
, J
t+1
,) in a sequence, the image to be processed being the central image (J
t
) of the sequence, which method also includes the following steps: testing of said three temporal intensities (G
t−1
, G
t
, G
t+1
), previously smoothed for noise peaks, in order to distinguish a first case in which the smoothed temporal intensity (G
t
) in the central image (J
t
) is substantially different from the other two smoothed temporal intensities (G
t−1
, G
t+1
) from a second case in which the smoothed temporal intensity (G
t
) in the central image (J
t
) is substantially equal to at least one of the other two smoothed temporal intensities (G
t−1
, G
t+1
), and assignment of a noise filtered intensity (R
t
) to the current pixel in the central image (J
t
) which in the first case is directly the smoothed temporal intensity (G
t
) at the current pixel in the central image and which in the second case is the result of a median filter applied to the three smoothed temporal intensities (G
t−1
, G
t
, G
t+1
) in the second case.
An advantage of the method consists in that it is not very complex: it is merely necessary to define one parameter, being the number of pixels in the elementary mask of the filter FMH and the threshold.
A medical imaging apparatus which includes means for carrying out this method is defined in claim
8
. In the fluoroscopy mode this apparatus enables on-line tracking of very delicate operations, utilizing very small tools, with an image quality and preservation of the movement of the small tools which could otherwise be achieved only by means of a much higher radiation dose which would be detrimental to the patient and to the radiologist.
REFERENCES:
patent: 5909515 (1999-06-01), Makram-Ebeid
patent: 5911012 (1999-06-01), Bernard et al.
patent: 6037986 (2000-03-01), Zhang et al.
patent: 6151417 (2000-11-01), Florent
patent: 6154519 (2000-11-01), Florent et al.
patent: 6281942 (2001-08-01), Wang
patent: 0 512 422 (1992-11-01), None
patent: 0 578 311 (1994-01-01), None
“A system for the removal of Impulsive Noise in image sequences”, Kokaram A. C. et al, Proceedings of the SPIE, vol. 1818, No. 1, Jan. 1, 1992, pp. 322-331.*
“A new class of Detail-Preserving Filters for Image Processing” by Ari Nieminen, Poekka Heinonen, and Yrjo Neuvo in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-9, No. 1, Jan. 1987.
Rogers Scott
U.S. Philips Corporation
Vodopia John
LandOfFree
Image processing method for reducing the noise in an image... 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 for reducing the noise in an image..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing method for reducing the noise in an image... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2849092