Method and circuit for detecting edge of fixed pattern

• US 8,594,436 B2
• Filed: 06/07/2012
• Issued: 11/26/2013
• Est. Priority Date: 11/25/2011
• Status: Expired due to Fees

• Current Assignee: Novatek Microelectronics Corporation (Novatek Microelectronics Corp Limited)
• Original Assignee: Novatek Microelectronics Corporation (Novatek Microelectronics Corp Limited)
• Inventors: Chen, Chun-Wei, Jiang, Jiande, Chen, Tsui-Chin, Liu, Yu-Shu
• Primary Examiner(s): Patel, Jayesh A
• Application Number:
  US13/490,476
• Publication Number:
  US 20130136365A1
• Time in Patent Office: 537 Days
• Field of Search: 382/199, 382170-172, 382260-266, 382/269, 348/625, 348/142, 348/208.4
• US Class Current: 382/199
• CPC Class Codes: G06V 10/752 Contour matching

Abstract

Method for detecting edge of fixed pattern includes receiving and analyzing a first image to obtain a first edge information. Second image and a corresponding second edge information are received, in which the second image includes an accumulation of image history information. According to the first edge information and the second edge information, a consistent number and an inconsistent number for pairs of pixels at the corresponding location of the first image and the second image are calculated, in which the consistent number represents how many pairs of pixels of which two compared pixels of each pair are both edge pixels, and the inconsistent number represents one of the two compared pixels is not the edge pixel. When the consistent number is greater than first predetermined value and meanwhile the inconsistent number is less than second predetermined value, first image and second image have a fixed pattern with fixed edge.

Claims

1. A method for detecting an edge of a fixed pattern, comprising:

receiving a current input image;

receiving a history image and a corresponding history edge information;

adding the current input image and the history image according to a weight ratio to obtain a current analyzing image;

calculating an edge strength value of each pixel of the current analyzing image, wherein the edge strength value is calculated based on a predetermined condition of edge characteristic, and when the edge strength value is greater than a predetermined value, the pixel is a possible edge pixel, the current analyzing image is taken as the history image for next edge detection and analysis, and the edge pixel and the edge strength are recorded in the history edge information;

calculating a consistent pixel number that meets a predetermined consistent condition and an inconsistent pixel number that meets a predetermined inconsistent condition for the edge pixel of the current analyzing image and the edge pixel of the history image; and

when the consistent pixel number is greater than a first predetermined value and meanwhile the inconsistent pixel number is less than a second predetermined value, detecting the current input image having a fixed pattern with a fixed edge.

2. The method for detecting an edge of a fixed pattern according to claim 1, wherein the edge characteristic condition comprises:

calculating four pixel value differences of the four pairs of pixels respectively adjacent to an analyzing pixel and extending in a horizontal direction, a vertical direction, a first oblique direction and a second oblique direction by taking the analyzing pixel as a center,

wherein when at least one of four pixel absolute difference values is greater than a threshold, the analyzing pixel is one of the edge pixels, and the largest one of the four pixel absolute difference values is the edge strength value of the edge pixel.

3. The method for detecting an edge of a fixed pattern according to claim 1, wherein the consistent condition comprises:

for two compared pixels at the corresponding location of the current analyzing image and the history image, if the two compared pixels are both the edge pixels, and respectively an absolute difference of the edge strength value is less than a predetermined value, the two compared pixels are consistent, and the consistent number is added with one, wherein the absolute difference of the edge strength value comprises considering whether the two compared pixels are in the same direction of the edge and whether a slope of the edge strength value has the similar condition.

4. The method for detecting an edge of a fixed pattern according to claim 3, wherein the inconsistent condition comprises:

if one of the two compared pixels is the edge and the other is not the edge, the two compared pixels are inconsistent and the inconsistent number is added with one.

5. The method for detecting an edge of a fixed pattern according to claim 1, wherein the weight ratio corresponding to the current analyzing image and the history image is a:

(1−

a), where a is an adjustable value in a range of 0 to 1.

6. The method for detecting an edge of a fixed pattern according to claim 1, wherein the edge detecting for the current analyzing image is carried out with a whole piece of the image or a block as a detecting unit.

7. A method for detecting an edge of a fixed pattern, comprising:

receiving and analyzing a first image to obtain a first edge information;

receiving a second image and a corresponding second edge information, in which the second image comprises an accumulation of image history information;

according to the first edge information and the second edge information, calculating a consistent number and an inconsistent number for pairs of pixels at the corresponding location of the first image and the second image, wherein the consistent number represents how many pairs of pixels of which two compared pixels of each pair are both edge pixels, and the inconsistent number represents one of the two compared pixels is not the edge pixel; and

when the consistent number is greater than a first predetermined value and meanwhile the inconsistent number is less than a second predetermined value, the first image and the second image have a fixed pattern with a fixed edge.

8. The method for detecting an edge of a fixed pattern according to claim 7, wherein the condition for determining whether a pixel is the edge pixel comprises:

calculating four pixel value absolute differences of the four pairs of pixels respectively adjacent to an analyzing pixel and extending in a horizontal direction, a vertical direction, a first oblique direction and a second oblique direction by taking the analyzing pixel as a center,

wherein when at least one of the four pixel absolute differences is greater than a threshold, the analyzing pixel has an edge pixel attribute.

9. The method for detecting an edge of a fixed pattern according to claim 7, wherein when the two compared pixels are both the edge pixels, the statistic of the consistent number further meets the following conditions:

the two compared pixels respectively have two edge strength values and the two edge strength values have an absolute difference that is less than a predetermined value.

10. The method for detecting an edge of a fixed pattern according to claim 7, wherein the edge detecting for the current analyzing image is carried out with a whole piece of the image or a block as a detecting unit.

11. A circuit for detecting an edge of a fixed pattern, comprising:

an image overlapping unit, for receiving and storing a current input image and a history image and a corresponding history edge information, wherein the current input image and the history image are added according to a weight ratio to obtain a current analyzing image;

an edge detecting unit, for calculating an edge strength value of each pixel for the current analyzing image, wherein the edge strength value is calculated based on a predetermined condition of edge characteristic, and when the edge strength value is greater than a predetermined value, the pixel is a possible edge pixel, wherein the current analyzing image is taken as the history image for next edge detection and analysis, and the edge pixel and the edge strength are recorded in the history edge information; and

a fixed pattern detecting unit, for calculating a consistent pixel number that meets a predetermined consistent condition and an inconsistent pixel number that meets a predetermined inconsistent condition for the edge pixel of the current analyzing image and the edge pixel of the history image,

wherein when the consistent pixel number is greater than a first predetermined value and meanwhile the inconsistent pixel number is less than a second predetermined value, the current input image is detected to have a fixed pattern with a fixed edge.

12. The circuit for detecting an edge of a fixed pattern according to claim 11, wherein the edge characteristic condition comprises:

calculating four pixel value absolute differences of the four pairs of pixels respectively adjacent to an analyzing pixel and extending in a horizontal direction, a vertical direction, a first oblique direction and a second oblique direction by taking the analyzing pixel as a center,

wherein when at least one of the four pixel absolute differences is greater than a threshold, the analyzing pixel is one of the edge pixels, and the largest one of the four pixel absolute difference values is the edge strength value of the edge pixel.

13. The circuit for detecting an edge of a fixed pattern according to claim 11, wherein the consistent condition comprises:

for two compared pixels at the corresponding location of the current analyzing image and the history image, if the two compared pixels are both the edge pixels, and respectively an absolute difference of the edge strength value is less than a predetermined value, the two compared pixels are consistent, and the consistent number is added with one.

14. The circuit for detecting an edge of a fixed pattern according to claim 13, wherein the inconsistent condition comprises:

if one of the two compared pixels is the edge and the other is not the edge, the two compared pixels are inconsistent and the inconsistent number is added with one.

15. The circuit for detecting an edge of a fixed pattern according to claim 11, wherein the weight ratio corresponding to the current analyzing image and the history image is a:

(1−

a), where a is an adjustable value in a range of greater than 0.5 and less than 1.