Patent EP0626656A2 successfully refreshed from Google Patents.
EP0626656A2
Expired / Lapsed
EP
⏱️ 0.0 years remaining
Image processing system and method for automatic feature extraction
Assignee: SOUTH BIRMINGHAM DISTRICT HEALTH AUTHORITY
•
Filing: 1994-04-28
•
Granted: —
•
Expires: 2014-04-28
Abstract
Abstract For extracting, from a source image, irregular relatively distinct features from an irregular relatively less distinct object which occupies a part of the source image, a number of image processing steps are performed in which: the edge outline of the object is extracted from the source image to produce an object outline image; the object outline is traced and predetermined data points on the object outline are located; particular regions of the object are identified within the source image using the identified data points; and the relatively distinct features are extracted from the identified regions. The invention finds particular application for the assessment of skeletal maturity ('bone age') wherein the source image is, or is derived from a digital or digitised radiograph of a subject's hand and wrist, with the object being the hand and wrist and the feature being the bones within the hand and/or wrist. In this case, classification of bone outlines extracted from the image are compared to and classified with respect to reference sample bones to compute an assessment of skeletal maturity.
Inventors
- 👤 Kevin Hill
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-11-07 22:06
CPC Classifications
G; G06; G06T; G06T7/00; G06T7/0002; G06T7/0012; G06T7/10; G06T7/12; G06T7/70; G06V; G06V10/00; G06V10/20; G06T2207/00; G06T2207/10; G06T2207/10116; G06T2207/30; G06T2207/30004; G06T2207/30008
Claims Analysis
Total Claims: 17
Claim 1
Independent Claim
132 words
An image processing system for extracting, from a source image comprising an array of pixels, irregular relatively distinct features from an irregular relatively less distinct object which occupies a part of the source image, the system comprising: - object outline extracting means (S1, S2) for extracting the edge outline of the object in the source image to produce an object outline image; - data point locating means (S3) for tracing the extracted object outline and for locating predetermined data points on the object outline; - region identification means (S4) responsive to the located data points to identify particular regions of the object within the source image; and - feature identification means (S5) responsive to image data relating to the source image within the identified regions for extracting the relatively distinct features therefrom.
Claim 2
Dependent Claim
86 words
A system according to claim 1 wherein the object outline extracting means comprises: - object outline emphasising means (S2) including a difference function generator for replacing each pixel value in the image by a pixel value determined as a function of the difference in grey scale of pixels in a predetermined area centred on the pixel in question to produce an outline emphasised image; and - thresholding means (S2) for extracting, as the object outline, pixels from the outline emphasised image which exceed a predetermined threshold.
Claim 3
Dependent Claim
38 words
A system according to claim 2 wherein the thresholding means (S2) recursively determines a threshold on the basis of a cumulative distribution of grey scale levels in the outline emphasised image until a continuous outline has been extracted.
Claim 4
Dependent Claim
35 words
A system according to claim 2 or claim 3 wherein the object outline emphasising means also comprises a filter (S1) for reducing the noise in the source image before processing by the difference function generator.
Claim 5
Independent Claim
49 words
A system according to any one of claims 2 to 4 wherein the object outline extracting means (S1, S2) removes any object pixels forming a part of the object outline which part has a perimeter greater than the distance between its extremities by an amount outside a predetermined range.
Claim 6
Independent Claim
62 words
A system according to any one of the preceding claims wherein the data point locating means (S3) comprises: - means for determining the orientation of the object within the image and initial data points by identifying predetermined object structures within the image; and - means for selecting further data points corresponding to predetermined directions or changes in direction of the object outline.
Claim 7
Independent Claim
73 words
A system according to any one of the preceding claims wherein the region identification means (S4) comprises sub-image defining means responsive to the data point locating means define a sub-image identified by respective data points by extracting pixels from or derived from the source image and, where necessary, rotating the pixels to correspond to a predetermined sub-image orientation, and wherein the feature identification means identifies the outlines of individual features within a sub-image.
Claim 8
Independent Claim
39 words
A system according to any one of the preceding claims wherein the feature identification means (S5) comprises: - means for emphasising the feature outlines; and - means for following the feature outlines to encode the shape of the outlines.
Claim 9
Independent Claim
57 words
A system according to any one of the preceding claims additionally comprising feature classifying means (28) which compares the features identified by the feature identification means to predetermined reference features to classify the identified features and report generating means for outputting a report on the result of the classification of predetermined identified features from the source image.
Claim 10
Independent Claim
49 words
A system according to any one of the preceding claims wherein the source image is, or is derived from a digital radiographic image of a part of a limb of a subject wherein the object is part of the limb and the features are the bones within the limb.
Claim 11
Dependent Claim
18 words
A system according to claim 10 wherein the part of a limb includes a hand of a subject.
Claim 12
Dependent Claim
65 words
A system according to claim 11 wherein the data point locating means (S3) comprises means for tracing fields, e.g. circles, of differing sizes, e.g. diameters, on the object outline image and for determining data points representative of the wrist or forearm, as start and finish points for subsequently tracing the object outline, in response to the detection of a predetermined pixel value pattern or patterns.
Claim 13
Dependent Claim
71 words
A system according to claim 11 or claim 12 when dependent on claim 7 wherein the feature identification means (S5) comprises means for assessing the relative widths of the relatively distinct areas in the wrist portion of source image, or of an image derived therefrom, to determine the handedness of the image, and, where necessary, to invert the pixels of the sub-images to correspond to a hand of a particular handedness.
Claim 14
Dependent Claim
26 words
A system according to any one of claims 10 to 13 when dependent on claim 9 wherein the report generated is an assessment of skeletal maturity.
Claim 15
Dependent Claim
57 words
A system according to claim 14 wherein the feature classifying means (28) comprises rough age assessment means for scanning pixels within an identified image region to identify predetermined pixel value changes indicative of immature bone development whereby a rough age assessment is computed for distinguishing different cases relevant to the generation of the report on skeletal maturity.
Claim 16
Independent Claim
47 words
A system according to any one of the preceding claims comprising means (24) for generating the source image from an initial image by adjusting the spread of pixel values of the initial image to utilise substantially the full range of pixel values available for the source image.
Claim 17
Independent Claim
112 words
A method for processing an image to extract, from a source image comprising an array of pixels, irregular relatively distinct features from an irregular relatively less distinct object which occupies a part of the source image, the method comprising: a) extracting an edge outline of the object in the source image to produce an object outline image; b) tracing the extracted object outline and locating predetermined data points on the object outline; c) responding to the located data points to identify particular regions of the object within the source image; and d) responding to image data relating to the source image within the identified regions for extracting the relatively distinct features therefrom.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.