Image correction method and related image correction system thereof

• US 9,153,014 B2
• Filed: 04/21/2011
• Issued: 10/06/2015
• Est. Priority Date: 11/09/2010
• Status: Active Grant

• Current Assignee: Avisonic Technology Corporation (Elan Microelectronics Corporation)
• Original Assignee: Avisonic Technology Corporation (Elan Microelectronics Corporation)
• Inventors: Yu, Chi-Chang, Chen, Homer H., Lin, Meng-Chun
• Primary Examiner(s): Thirugnanam, Gandhi
• Application Number:
  US13/091,137
• Publication Number:
  US 20120114262A1
• Time in Patent Office: 1,629 Days
• Field of Search: 382/103, 382/254
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 5/80 Geometric correction

Abstract

The present invention provides an image correction method and a related image correction system which can correct images captured via a fisheye lens or a ultra-wide angle lens camera so as to alleviate geometrical distortion in the images, and geometrically adjust the images according to user'"'"'s requirements. Wherein, the present invention further enhances the processing performance of image correction computation by a memory allocation technique.

Claims

1. An image correction method, for outputting a corrected image according to an image to be corrected, the corrected image comprising a plurality of corrected pixel blocks, the image to be corrected being captured by a fisheye lens camera, the image correction method performing a plurality of pixel value computations, each pixel value computation being utilized for generating a pixel value of each corrected pixel within a corrected pixel block, and comprising:

establishing a coordinate transformation relation between the image to be corrected and the corrected image, such that a first coordinate value of each corrected pixel within the corrected pixel block corresponds to a second coordinate value within the image to be corrected, wherein the coordinate transformation relation is determined by at least a perspective distortion correction coordinate transformation, and in the perspective correction distortion coordinate transformation, an angle parameter corresponding to an object plane and an image plane is determined by an incidence angle of an incident beam corresponding to the image to be corrected, where the incident beam passes through an optical center of the fisheye lens;

according to the coordinate transformation relation, determining a pixel block to be corrected including a plurality of pixels, which corresponds to the corrected pixel block, and selecting a pixel block to be processed which comprises at least the pixel block to be corrected from the image to be corrected; and

utilizing the second coordinate value corresponding to the corrected pixel for determining a plurality of specific pixels to be processed that correspond to the corrected pixel, and generating a pixel value of the corrected pixel by performing an interpolation computation upon a plurality of pixel values of the specific pixels to be processed.

2. The image correction method of claim 1, further comprising:

temporarily storing a pixel value of each pixel to be processed into a memory device having a plurality of memory banks, wherein pixel values of pixels to be processed that are located at different rows are temporarily stored into 20different memory banks of the memory device, respectively;

and the step of generating the pixel value of the corrected pixel comprises;

reading the plurality of pixels values of the specific pixels to be processed from the memory device;

wherein the pixel values of the specific pixels to be processed are temporarily stored in different memory banks.

3. The image correction method of claim 2, wherein a number of the specific pixels to be processed is m*n, a number of the memory banks within the memory device is m, grabs continuous n pixels from each memory bank with the same memory address offset to get m by n pixels in one cycle time, and m and n are both positive integers.

4. An image correction system, for outputting a corrected image according to an image to be corrected, the corrected image comprising a plurality of corrected pixel blocks, the image to be corrected being captured by a fisheye lens camera, the image correction system performing a plurality of pixel value computations, each pixel value computation being utilized for generating a pixel value of each corrected pixel within the corrected pixel block, the image correction system comprising:

a coordinate transformation relation generating circuit, for establishing a coordinate transformation relation between the image to be corrected and the corrected image, such that a first coordinate value of each corrected pixel within the corrected pixel block corresponds to a second coordinate value within the image to be corrected, wherein the coordinate transformation relation generating circuit comprises at least a perspective distortion correction coordinate transformation circuit, and the coordinate transformation relation is determined by at least the perspective distortion correction coordinate transformation circuit, and an angle parameter utilized by the perspective distortion correction coordinate transformation circuit and corresponding to an object plane and an image plane is determined by an incidence angle of an incident beam corresponding to the image to be corrected, where the incident beam passes through an optical center of the fisheye lens;

a block selecting circuit, coupled to the coordinate transformation relation generating circuit, for referring to the coordinate transformation relation for determining a pixel block to be corrected including a plurality of pixels, which corresponds to the corrected pixel block, and selecting a pixel block to be processed which comprises at least the pixel block to be corrected from the image to be corrected; and

an interpolation computation circuit, coupled to the coordinate transformation relation generating circuit, the interpolation computation circuit utilizing the second coordinate value corresponding to the corrected pixel for determining a plurality of specific pixels to be processed that correspond to the corrected pixel, and generating a pixel value of the corrected pixel by performing an interpolation computation upon a plurality of pixel values of the specific pixels to be processed.

5. The image correction system of claim 4, further comprising:

a memory device, coupled to the block selecting circuit, the memory device having a plurality of memory banks for temporarily storing a pixel value of each pixel to be processed;

wherein the interpolation computation circuit reads the plurality of pixel values of the specific pixels to be processed from the memory device, and the pixel values of the specific pixels to be processed are temporarily stored in different memory banks.

6. The image correction system of claim 5, wherein a number of the specific pixels to be processed is m*n, a number of the memory banks within the memory device is m, grabs continuous n pixels from each memory bank with the same memory address offset to get m by n pixels in one cycle time and m and n are both positive integers.