EDGE NEIGHBORHOOD DESCRIPTOR

• US 20110235922A1
• Filed: 07/27/2009
• Published: 09/29/2011
• Est. Priority Date: 07/29/2008
• Status: Active Grant

• Current Assignee: Superfish Incorporated
• Original Assignee: Superfish Incorporated
• Inventors: Chertok, Michael, Koushnir, Lev, Pinhas, Adi
• Granted Patent: US 8,463,047 B2
• Time in Patent Office: Days
• US Class Current: 382/199
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]

Abstract

System for generating an edge neighborhood descriptor for describing the surrounding of an interest point according to the closest edges includes a sector determiner, a closest edge determiner, and an edge neighborhood descriptor constructor, the closest edge determiner is coupled between the sector determiner and the edge neighborhood descriptor constructor. The sector determiner determines N sectors, surrounding the interest point. The closest edge determiner determines for each of the N sectors, the edge pixel closest to the interest point, according to at least one binary edge map. The edge neighborhood descriptor constructor constructs the edge neighborhood descriptor such that the length of the radius of each of the N sectors is determined according to at least the distance from the interest point to the edge pixel closest to the interest point within the sector, the edge neighborhood descriptor includes the N sectors.

Claims

1. System for generating an edge neighborhood descriptor for describing the surrounding of an interest point according to the closest edges, the system comprising:

a sector determiner for determining N sectors, surrounding said interest point;

a closest edge determiner, coupled with said sector determiner, for determining for each of said N sectors, the edge pixel closest to said interest point, according to at least one binary edge map;

an edge neighborhood descriptor constructor coupled with said closest edge determiner, for constructing said edge neighborhood descriptor such that the length of the radius of each of said N sectors is determined according to at least the distance from said interest point to the edge pixel closest to said interest point within said sector, wherein said edge neighborhood descriptor includes said N sectors.

2. The system of claim 1, further comprising an interest point detector coupled with said sector determiner, said interest point detector receiving an image of an object and detecting said interest point within said image.

3. The system of claim 1, further comprising an edge detector coupled with said sector determiner, said edge detector receiving an image of an object and producing said at least one binary edge map.

4. The system of claim 1, wherein said sector determiner determines N equal sectors, the radial angle of each of said equal sectors conforming to an angle of 2π

IN.

5. The system of claim 1, wherein said sector determiner determines N sectors, the radial angle of at least one of said sectors conforms to a different angle than the radial angle of at least another one of said sectors.

6. The system according to claim 1, further comprising a distance normalizer coupled with said closest edge determiner and with said edge neighborhood descriptor constructor, said distance normalizer determines the length of the radius of a selected one of said sectors as a normalization marker, said distance normalizer normalizing the length of the radius of each of said N sectors according to said normalization marker.

7. The system according to claim 1, further comprising an edge neighborhood descriptor orientator coupled with said closest edge determiner and with said edge neighborhood descriptor constructor, said edge neighbor hood descriptor orientator determining the orientation of a selected one of said sectors as an orientation marker, said edge neighbor hood descriptor orientator orientating said edge neighborhood descriptor according to said orientation marker.

8. Method for generating an edge neighborhood descriptor for describing the surrounding of an interest point according to the closest edges, the method comprising the procedures of:

dividing the surrounding of said interest point into N sectors;

for each of said N sectors, determining the closest edge pixel to said interest point according to at least one binary edge map; and

constructing said edge neighborhood descriptor such that the length of the radius of each of said N sectors is determined according to at least the respective one of an N distances distance from the closest edge pixel in each of said N sectors to said interest point.

9. The method of claim 8, further comprising the procedure of detecting said interest point within an image of an object, performed prior to said procedure of dividing.

10. The method of claim 8, further comprising the procedure of constructing said at least one binary edge map, performed prior to said procedure of dividing.

11. The method of claim 8, wherein said N sectors are equal, the radial angle of each of said equal sectors conforming to an angle of 2π

/N.

12. The method of claim 8, wherein the radial angle of at least one of said N sectors conforms to a different angle that the radial angle of at least another one of said N sectors.

13. The method according to claim 8, further comprising the procedures of:

determining a normalization marker according to the length of a selected one of said N sectors, and normalizing the values of lengths of each of said N sectors according to said normalization marker.

14. The method according to claim 8, further comprising the procedured of:

determining an orientation marker according to the orientation of a selected one of said N sectors, and orientating said edge neighborhood descriptor according to said normalization marker.