Image stabilization method

• US 7,746,934 B2
• Filed: 04/19/2006
• Issued: 06/29/2010
• Est. Priority Date: 04/19/2006
• Status: Active Grant

• Current Assignee: Imagination Technologies Limited (Imagination Technologies Group Ltd.)
• Original Assignee: Nethra Imaging Inc. (Renesas Electronics Corporation)
• Inventors: Chen, Hong, Wong, Ping Wah
• Primary Examiner(s): WONG, ALLEN C
• Application Number:
  US11/407,450
• Publication Number:
  US 20070248166A1
• Time in Patent Office: 1,532 Days
• Field of Search: 375/240, 375/240.01, 375/240.25, 375/240.29, 375/240.12, 375/240.16, 348/208.8, 348/223.1, 348/470, 348/22, 382/261, 382/233
• US Class Current: 375/240.29
• CPC Class Codes: H04N 19/44 Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder
• CPC Class Codes: H04N 19/51 Motion estimation or motion compensation
• CPC Class Codes: H04N 19/80 Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation H04N19/635, H04N19/86 take precedence
• CPC Class Codes: H04N 19/86 involving reduction of coding artifacts, e.g. of blockiness

Abstract

In one embodiment, a method for reducing motion artifacts in an output image is provided. The method comprises capturing multiple frames of a scene; determining local motion vectors between each pixel in a current frame and a corresponding pixel in the a previous frame; and performing a temporal filtering operation based on the local motion vectors wherein pixels from a plurality frames are integrated to form the output image.

Claims

1. A method for reducing motion artifacts in an output image, comprising:

capturing multiple frames of a scene;

determining local motion vectors between each pixel in a current frame and a corresponding pixel in a previous frame; and

performing a selective temporal filtering operation based on the local motion vectors to form the output image;

wherein said temporal filtering operation (a) rejects a local motion vector if the local motion vector is above a threshold in which case no temporal filtering is performed;

(b) averages a current pixel with the corresponding pixel from the previous frame if the local motion vector for the current pixel is considered zero; and

(c) averages the current pixel with the motion-compensated pixel from the previous frame in all other cases.

2. The method of claim 1, wherein an effective exposure time for the output image is longer than for each of the multiple frames.

3. The method of claim 1, wherein determining the local motion vectors comprises for each pixel location (m,n) in the current frame, defining a block of pixels centered at (m,n) in the current frame and finding a block in the previous frame that is the closest match to the block of pixels centered at (m,n) in the current frame.

4. The method of claim 3, further comprising applying a smoothing filter to the current and previous frames prior to calculating the local motion vectors.

5. The method of claim 3, wherein the smoothing filter comprises a spatial low pass filter.

6. The method of claim 3, wherein performing the temporal filtering operation comprises selectively adjusting for pixel motion between a current and a previous frame.

7. The method of claim 6, wherein selectively adjusting for pixel motion comprises rejecting a motion vector if a degree of similarity between the closest matching block in the previous frame to the block of pixels in the current frame centered on (m,n) exceeds a predefined block rejection threshold.

8. The method of claim 6, wherein selectively adjusting for pixel motion comprises applying a motion vector of zero if a degree of similarity between the closest matching block in the previous frame to the block of pixels in the current frame centered on (m,n) is below a predefined block noise threshold.

9. The method of claim 1, wherein in determining the closest match, luminance values for the pixels in each block are compared.

10. The method of claim 1, wherein the temporal filtering operation comprises applying a finite impulse response filter.

11. The method of claim 1, wherein the temporal filtering operation comprises applying an infinite impulse response filter.

12. An image processor, comprising:

an image buffer to store image data for a captured image; and

image stabilization logic to reduce motion artifacts in an output image, wherein the image stabilization logic captures multiple frames of a scene;

determines local motion vectors between each pixel in a current frame and a corresponding pixel in a previous frame; and

performs a selective temporal filtering operation based on the local motion vectors to form the output image;

wherein said temporal filtering operation (a) rejects a local motion vector if the is above a threshold in which case no temporal filtering is performed;

(b) averages a current pixel with the corresponding pixel from the previous frame if the local motion vector for the current pixel is considered zero; and

(c) averages the current pixel with its motion-compensated pixel from the previous frame in all other cases.

13. The image processor of claim 12, wherein determining the local motion vectors comprises for each pixel location (m,n) in the current frame, defining a block of pixels centered at (m,n) in the current frame and finding a block in the previous frame that is the closest match to the block of pixels centered at (m,n) in the current frame.

14. The image processor of claim 12, wherein the temporal filtering comprises applying either a finite impulse response filter or an infinite impulse response filter.

15. A camera system, comprising:

camera optics;

an image sensor positioned so that light passing through the camera optics impinges on the image sensor; and

an image processor coupled to the image sensor to receive image data for a captured image therefrom, wherein the image processor comprises image stabilization logic to perform a method for reducing motion artifacts in an output image, comprising;

capturing multiple frames of a scene;

determining local motion vectors between each pixel in a current frame and a corresponding pixel in the a previous frame; and

performing a temporal filtering operation based on the local motion vectors to form the output image;

wherein said temporal filtering operation (a) rejects a local motion vector if the local motion vector is above a threshold in which case no temporal filtering is performed;

(b) averages a current pixel with the corresponding pixel from the previous frame if the local motion vector for the current pixel is considered zero; and

(c) averages the current pixel with the motion-compensated pixel from the previous frame in all other cases.

16. The camera system of claim 15, wherein determining the local motion vectors comprises for each pixel location (m,n) in the current frame, defining a block of pixels centered at (m,n) in the current frame and finding a block in the previous frame that is the closest match to the block of pixels centered at (m,n) in the current frame.

17. The camera system of claim 15, wherein the temporal filtering operation comprises either a finite impulse response filter or an infinite impulse response filter.