| US 7,555,161 B2 | ||
| Image analysis | ||
| John Francis Haddon, Farnborough (United Kingdom); and Sharon Katrina Watson, Farnborough (United Kingdom) | ||
| Assigned to Qinetiq Limited, (United Kingdom) | ||
| Appl. No. 10/533,204 PCT Filed Nov. 10, 2003, PCT No. PCT/GB03/04841 § 371(c)(1), (2), (4) Date Apr. 28, 2005, PCT Pub. No. WO2004/044845, PCT Pub. Date May 27, 2004. |
||
| Claims priority of application No. 0226331.7 (GB), filed on Nov. 12, 2002. | ||
| Prior Publication US 2006/0002608 A1, Jan. 05, 2006 | ||
| Int. Cl. G06K 9/34 (2006.01) | ||
| U.S. Cl. 382—173 [381/168; 381/171] | 19 Claims |
| 1. A method for the automated analysis of a digital image comprising an array of pixels, including the steps of using a computer
or processor for:
identifying certain pixels to exclude from the following step, consequent upon a local property of such pixels;
generating a property co-occurrence matrix (PCM) from pixels not identified to be excluded by the preceding step, using the
properties of local mean and local standard deviation of intensity in neighbourhoods of the included pixels; and
segmenting the image by labelling the pixels which are included in the formation of said PCM as belonging to specified classes
consequent upon analysis of said PCM and by labeling pixels which are excluded from the formation of said PCM as belonging
to a specified class different from the first-mentioned classes.
|