Image matching method and stereo matching system

• US 9,042,638 B2
• Filed: 09/22/2013
• Issued: 05/26/2015
• Est. Priority Date: 07/04/2013
• Status: Active Grant

• Current Assignee: Novatek Microelectronics Corporation (Novatek Microelectronics Corp Limited)
• Original Assignee: Novatek Microelectronics Corporation (Novatek Microelectronics Corp Limited)
• Inventors: Zhou, Lei, Liu, Guangzhi, Chou, Cheng-Wei
• Primary Examiner(s): Chu, Randolph I
• Application Number:
  US14/033,515
• Publication Number:
  US 20150010230A1
• Time in Patent Office: 611 Days
• Field of Search: None
• US Class Current: 382/154
• CPC Class Codes: G06F 2218/12 Classification; Matching
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 7/593 from stereo images
• CPC Class Codes: G06V 10/751 Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

An image matching method is utilized for performing a stereo matching from a first image block to a second image block in a stereo matching system. The image matching method includes performing a matching computation from the first image block to the second image block according to a first matching algorithm to generate a first matching result; performing the matching computation between the first image block and the second image block according to a second matching algorithm to generate a second matching result and a third matching result; obtaining a matching error and a matching similarity of the first image block according to the second matching result and the third matching result; and determining a stereo matching result of the first image block according to the matching error and the matching similarity.

Claims

1. An image matching method, for performing a stereo matching from a first image block to a second image block in a stereo matching system, the image matching method comprising:

performing a matching computation from the first image block to the second image block according to a first matching algorithm to generate a first matching result;

performing the matching computation from the first image block to the second image block according to a second matching algorithm to generate a second matching result;

performing the matching computation from the second image block to the first image block according to the second matching algorithm to generate a third matching result;

obtaining a matching error of the first image block corresponding to the second matching algorithm according to the second matching result and the third matching result, and obtaining a matching similarity of the first image block matched to the second image block according to the second matching result; and

determining a stereo matching result of the first image block as the first matching result or the second matching result according to the matching error and the matching similarity.

2. The image matching method of claim 1, wherein the first image block and the second image block are two scan lines or two regions of an image.

3. The image matching method of claim 1, wherein the first matching algorithm is designed according to a sequential correlation existing between the first image block and the second image block, and the first matching algorithm is adapted to the matching computation of the first image block and the second image block having a large object.

4. The image matching method of claim 1, wherein the first matching algorithm performs the matching computation according to a dynamic programming algorithm.

5. The image matching method of claim 1, wherein the second matching algorithm is designed according to a sequential correlation not existing between the first image block and the second image block, and the second matching algorithm is adapted to the matching computation of the first image block and the second image block having a small object.

6. The image matching method of claim 1, wherein the second matching algorithm performs the matching computation according to a scan-line optimization algorithm.

7. The image matching method of claim 1, wherein the step of obtaining the matching error of the first image block corresponding to the second matching algorithm according to the second matching result and the third matching result comprises:

obtaining a matching point of the second image block matched from a representative point of the first image block according to the second matching result;

obtaining a verification point of the first image block matched from the matching point of the second image block according to the third matching result; and

calculating a distance difference between the representative point and the verification point to obtain the matching error of the first image block corresponding to the second matching algorithm.

8. The image matching method of claim 1, wherein the step of obtaining the matching similarity of the first image block matched to the second image block according to the second matching result comprises:

obtaining at least one similar point of the second image block matched from at least one original point of the first image block according to the second matching result; and

obtaining the matching similarity according to the at least one original point and the at least one similar point.

9. The image matching method of claim 8, wherein the matching similarity is a sum of at least one absolute pixel value difference between the at least one original point and the at least one similar point.

10. The image matching method of claim 1, wherein the step of determining the stereo matching result of the first image block as the first matching result or the second matching result according to the matching error and the matching similarity comprises:

determining the stereo matching result of the first image block as the second matching result when the matching error is less than an error threshold and the matching similarity is less than a similarity threshold, and determining the stereo matching result of the first image block as the first matching result when the matching error is greater than the error threshold or the matching similarity is greater than the similarity threshold.

11. The image matching method of claim 1, wherein the stereo matching result includes disparity information of the first image block matched to the second image block.

12. A stereo matching system, for performing a stereo matching from a first image block to a second image block, the stereo matching system comprising:

a first matching module for performing a matching computation from the first image block to the second image block according to a first matching algorithm to generate a first matching result;

a second matching module for performing the matching computation from the first image block to the second image block according to a second matching algorithm to generate a second matching result, performing the matching computation from the second image block to the first image block according to the second matching algorithm to generate a third matching result, obtaining a matching error of the first image block corresponding to the second matching algorithm according to the second matching result and the third matching result, and obtaining a matching similarity of the first image block matched to the second image block according to the second matching result; and

a determination module for determining a stereo matching result of the first image block as the first matching result or the second matching result according to the matching error and the matching similarity.

13. The stereo matching system of claim 12, wherein the first image block and the second image block are two scan lines or two regions of an image.

14. The stereo matching system of claim 12, wherein the first matching algorithm is designed according to a sequential correlation existing between the first image block and the second image block, and the first matching algorithm is adapted to the matching computation of the first image block and the second image block having a large object.

15. The stereo matching system of claim 12, wherein the first matching algorithm performs the matching computation according to a dynamic programming algorithm.

16. The stereo matching system of claim 12, wherein the second matching algorithm is designed according to a sequential correlation not existing between the first image block and the second image block, and the second matching algorithm is adapted to the matching computation of the first image block and the second image block having a small object.

17. The stereo matching system of claim 12, wherein the second matching algorithm performs the matching computation according to a scan-line optimization algorithm.

18. The stereo matching system of claim 12, the step of obtaining the matching error of the first image block corresponding to the second matching algorithm according to the second matching result and the third matching result comprises:

obtaining a matching point of the second image block matched from a representative point of the first image block according to the second matching result;

obtaining a verification point of the first image block matched from the matching point of the second image block according to the third matching result; and

calculating a distance difference between the representative point and the verification point to obtain the matching error of the first image block corresponding to the second matching algorithm.

19. The stereo matching system of claim 12, wherein the step of obtaining the matching similarity of the first image block matched to the second image block according to the second matching result comprises:

obtaining at least one similar point of the second image block matched from at least one original point of the first image block according to the second matching result; and

obtaining the matching similarity according to the at least one original point and the at least one similar point.

20. The stereo matching system of claim 19, wherein the matching similarity is a sum of at least one absolute pixel value difference between the at least one original point and the at least one similar point.

21. The stereo matching system of claim 12, wherein the step of determining the stereo matching result of the first image block as the first matching result or the second matching result according to the matching error and the matching similarity comprises:

determining the stereo matching result of the first image block as the second matching result when the matching error is less than an error threshold and the matching similarity is less than a similarity threshold, and determining the stereo matching result of the first image block as the first matching result when the matching error is greater than the error threshold or the matching similarity is greater than the similarity threshold.

22. The stereo matching system of claim 12, wherein the stereo matching result includes disparity information of the first image block matched to the second image block.