IMAGE MATCHING APPARATUS AND METHOD

• US 20100142828A1
• Filed: 05/29/2009
• Published: 06/10/2010
• Est. Priority Date: 12/10/2008
• Status: Abandoned Application

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: Cho, Jae Il, Hwang, Dae Hwan, Choi, Seung Min, Chang, Ji Ho
• Application Number:
  US12/474,848
• Publication Number:
  US 20100142828A1
• Time in Patent Office: Days
• US Class Current: 382/209
• CPC Class Codes: G06T 7/593 from stereo images

Abstract

Provided are an image matching apparatus and method, and the image matching apparatus includes a determining unit determining whether a node, in which a first pixel of a left image of a subject and a second pixel of a right image of the subject corresponding to the first pixel are calculated, is a matchable region, and an operating unit calculating a disparity value by using the brightness information of a left window composed of the first pixel corresponding to the node and peripheral pixels surrounding the first pixel and the brightness information of a right window composed of the second pixel corresponding to the node and peripheral pixels surrounding the second pixel, when the node is the matchable region as a result of the determination.

Claims

1. An image matching apparatus, comprising:

a determining unit determining whether a node, in which a first pixel of a left image of a subject and a second pixel of a right image of the subject corresponding to the first pixel are calculated, is a matchable region; and

an operating unit calculating a disparity value by using the brightness information of a left window composed of the first pixel corresponding to the node and peripheral pixels surrounding the first pixel and the brightness information of a right window composed of the second pixel corresponding to the node and peripheral pixels surrounding the second pixel, when the node is the matchable region as a result of the determination.

2. The image matching apparatus of claim 1, wherein the operating unit calculates the disparity value by using the pixels peripheral to the first pixel and the pixels peripheral to the second pixel, when the node is an unmatchable region as a result of the determination.

3. The image matching apparatus of claim 2, wherein the operating unit comprises:

a first operator performing Sum of Absolute Difference (SAD) and Received Mean Census Transform (RMCT) on the left and right windows to calculate an energy value of a corresponding node corresponding to the first or second pixel in an synthesis image of the left and right images;

a second operator calculating an energy value of the corresponding node by using energy values of above and below nodes of the first pixel and energy values of above and below nodes of the second pixel; and

a third operator calculating the disparity value by using the energy value of the corresponding node.

4. The image matching apparatus of claim 3, wherein the first operator comprises:

a SAD processor performing SAD on the brightness information of the left and right windows;

a RMCT processor calculating average values of the brightness information of the left and right windows, performing Census Transform (CT) on the calculated average values, and outputting a hamming distance of the performed value; and

an adder adding an output of the SAD processor and an output of the RMCT processor,

wherein the first operator adds an output of the adder and an accumulated value of calculated energy values of nodes before the corresponding node.

5. The image matching apparatus of claim 4, further comprising a memory storing the accumulated value.

6. The image matching apparatus of claim 4, wherein the SAD processor comprises:

at least one of subtractor performing a subtraction operation on brightness information of respective nodes of the left window and brightness information of respective nodes of the right window corresponding to the respective nodes of the left window;

an absolute operator calculating absolute values of the respective subtraction results; and

at least one adder adding all the absolute values.

7. The image matching apparatus of claim 4, wherein the RMCT processor comprises:

at least one average value calculator outputting the average value of the brightness information of the left window and the average value of the brightness information of the right window;

a census transformer performing CT on the respective average values; and

a hamming distance operator comparing the CT results by bit to calculate a hamming distance.

8. The image matching apparatus of claim 3, wherein the second operator comprises a comparator comparing the energy value of the above node of the corresponding node with the energy value of the below node of the corresponding node to output a small value among the energy values.

9. An image matching apparatus, comprising:

a unit processing unit performing Sum of Absolute Difference (SAD) and Received Mean Census Transform (RMCT) on brightness information of left and right images to calculate an energy value of each node in an synthesis image of the left and right images;

a multi-processing unit calculating a matching value of a stereo image per each line by using the energy value of the each node; and

a rear processing unit calculating a disparity value of the stereo image by using the matching value.

10. The image matching apparatus of claim 9, wherein the unit processing unit accumulates a calculated energy value of a preceding node to the SAD and RMCT result per the each node to calculate the energy value of the each node.

11. The image matching apparatus of claim 9, wherein the unit processing unit performs SAD and RMCT on a left window composed of a first pixel of the left image corresponding to the each node and peripheral pixels surrounding the first pixel and a right window composed of a second pixel of the right image corresponding to the each node and peripheral pixels surrounding the second pixel to calculate the energy value of the each node.

12. The image matching apparatus of claim 9, wherein the unit processing unit calculates the energy value of the each node by using the SAD and RMCT result performed on the brightness information of the each node when the each node is a matchable region, and calculates the energy value of the each node by using a small value among energy values of above and below nodes of the each node when the each node is an unmatchable region.

13. The image matching apparatus of claim 12, wherein the above node and the below node are arranged about a disparity axis of the each node.

14. An image matching method, comprising:

determining whether a corresponding node, in which a first pixel of a left image of a subject and a second pixel of a right image of the subject corresponding to the first pixel are calculated, is a matchable region; and

calculating an energy value of the corresponding node by using the brightness information of a left window composed of the first pixel corresponding to a node and peripheral pixels surrounding the first pixel and the brightness information of a right window composed of the second pixel corresponding to the node and peripheral pixels surrounding the second pixel, when the first and second pixels are the matchable region as a result of the determination.

15. The image matching method of claim 14, further comprising calculating the energy value of the corresponding node by using energy values of pixels peripheral to the first and second pixels, when the first and second pixels are the unmatchable region as a result of the determination.

16. The image matching method of claim 15, wherein the determining of the matchable region comprises checking whether a sum of an order of a site axis of the first or second pixel and an order of a disparity axis of the first or second pixel is an odd or even number to determine the matchable region or the unmatchable region.

17. The image matching method of claim 14, wherein the calculating of the energy value comprises:

performing Sum of Absolute Difference (SAD) and Received Mean Census Transform (RMCT) on the left and right windows to calculate an energy value of a corresponding node corresponding to the first or second pixel in an synthesis image of the left and right images;

accumulating the calculated energy value of the corresponding node and energy values of nodes before the corresponding node to calculate a disparity value; and

configuring a matrix of the disparity value by line unit on the synthesis image.

18. The image matching method of claim 17, further comprising accumulating the energy values of the nodes before the corresponding node.

19. The image matching method of claim 17, wherein the calculating of the energy values comprises:

configuring the left window for the first pixel and the right window for the second pixel;

calculating absolute values of brightness information differences of the left and right windows by pixel unit, and adding the absolute values;

performing CT on an average value of brightness information of each pixel of the left and right windows to calculate a hamming distance; and

adding the added result and the calculated hamming distance to calculate an energy value of the corresponding node.