Image interpolation method and image interpolation system

Patent metadata

Field	Value
Publication number	US-9349158-B2
Application number	US-201414329991-A
Country	US
Kind code	B2
Filing date	Jul 14, 2014
Priority date	Jul 14, 2014
Publication date	May 24, 2016
Grant date	May 24, 2016

Abstract

An image interpolation method is utilized for performing an interpolation on a source image to obtain a destination image. The image interpolation method includes performing a domain transformation on a plurality of pixels of the source image to generate a plurality of first coefficients and a plurality of second coefficients; respectively determining an data interrelationship degree in at least one direction of each first coefficient to generate a plurality of direction results; performing a first interpolation process on the plurality of first coefficients according to the plurality of direction results to generate a plurality of first destination coefficients; performing a second interpolation process on the plurality of second coefficients to generate a plurality of second destination coefficients; performing a reverse domain transformation on the plurality of first destination coefficients and the plurality of second destination coefficients to obtain the destination image.

First claim

What is claimed is:

1. An image interpolation method, for performing an interpolation on a source image to obtain a destination image, comprising:
performing a domain transformation on a plurality of pixels of the source image, for generating a plurality of first coefficients and a plurality of second coefficients corresponding to the plurality of pixels;
analyzing and calculating a data correlation degree in at least one direction of each first coefficient according to a plurality of pixel locations of the plurality of pixels, respectively, for generating a plurality of direction results corresponding to the plurality of first coefficients;
performing a first interpolation process on the plurality of first coefficients according to the plurality of directional results, for generating a plurality of first destination coefficients;
performing a second interpolation process on the plurality of first second coefficients, for generating a plurality of second destination coefficients; and
performing an inverse domain transformation on the plurality of first destination coefficients and the plurality of second destination coefficients, for generating the plurality of destination pixels of the destination image.

2. The image interpolation method of claim 1 , wherein a domain transformation is a wavelet transformation.

3. The image interpolation method of claim 2 , wherein the plurality of first coefficients comprises high frequency components of the wavelet transformed results of the plurality pixels.

4. The image interpolation method of claim 1 , wherein the step of analyzing and calculating the data interrelationship degree in the at least one direction of each first coefficient according to the plurality of pixel locations of the plurality of pixels, respectively, for generating the plurality of direction results corresponding to the plurality of first coefficients comprises:
calculating at least one energy and at least one variance in the at least one direction of each first coefficient according to the plurality of pixel locations and the plurality of first coefficients;
analyzing the at least one energy and the at least one variance, for selecting a representative direction with the maximum data correlation degree in the at least one direction of each first coefficient, and generating a direction angle value corresponding to the representative direction;
acquiring a representative energy and a representative variance corresponding to the representative angle value from the at least one energy and the at least one variance and generating a direction confidence value of the representative direction according to the representative energy and the representative variance; and
generating each of the plurality of direction results with the direction angle value and the direction confidence value.

5. The image interpolation method of claim 4 , wherein the step of calculating the at least one energy and the at least one variance in the at least one direction of each first coefficient according to the plurality of pixel locations and the plurality of first coefficients comprises:
acquiring an analysis location corresponding to each first coefficient from the plurality of pixel locations;
acquiring coefficients passing through the analysis location and being on the at least one direction from the plurality of first coefficients according to the plurality of pixel locations, for generating at least one analysis data corresponding to the at least one direction;
performing operations on each of the at least one analysis data according to an energy algorithm, respectively, for generating the at least one energy corresponding to the at least one direction; and
performing operations on each of the at least one analysis data according to a variance algorithm, respectively, for generating the at least one variance corresponding to the at least one direction.

6. The interpolation method of claim 5 , wherein the step of acquiring coefficients passing through the analysis location and being on the at least one direction from the plurality of first coefficients according to the plurality of pixel locations, for generating the at least one analysis data corresponding to the at least one direction comprises:
acquiring a plurality of analysis coefficients passing through the analysis location and being on the analysis direction from the plurality of first coefficients according to the plurality of pixel locations; and
generating the analysis data comprising the plurality of analysis coefficients corresponding to the analysis direction.

7. The image interpolation method of claim 6 , wherein the step of performing the operations on each of the at least one analysis data according to the energy algorithm, respectively, for generating the at least one energy corresponding to the at least one direction comprises:
summing the absolute value or the square value of the plurality of the analysis coefficients of the analysis data, for generating a sum; and
dividing the sum by a number of the analysis coefficients, for generating the energy.

8. The image interpolation method of claim 6 , wherein the step of performing the operations on each of the at least one analysis data according to the variance algorithm, respectively, for generating the at least one variance corresponding to the at least one direction comprises:
averaging the plurality of analysis coefficients of the analysis data, for generating an average;
summing the absolute value or the square value of each of the plurality of analysis coefficients subtracts the average, for generating a sum; and
dividing the sum by a number of analysis coefficients, for generating the variance.

9. The image interpolation method of claim 4 , wherein the step of analyzing the at least one energy and the at least one variance, for selecting a representative direction with the maximum data correlation degree in the at least one direction of each first coefficient, and generating a direction angle value corresponding to the representative direction comprises:
analyzing the at least one energy, for acquiring the direction corresponding to a maximum energy among the at least one direction as the representative direction; or
analyzing the at least one variance, for acquiring the direction corresponding to a minimum variance among the at least one direction as the representative direction; or
performing operations on the at least one energy and the at least one variance, for generating at least one operation value, and analyzing the at least one operation value, for acquiring the direction with the operation value corresponding to a predefined value among the at least one direction as the representative direction.

10. The image interpolation method of claim 4 , wherein the step of generating the direction confidence value of the representative direction according to the representative energy and the representative variance comprises:
dividing the difference between the representative energy and representative variance by the representative energy, for generating a representative confidence value; and
limiting the representative confidence value to be within 0 and 1, for generating the direction confidence value.

11. The image interpolation method of claim 1 , wherein the first interpolation process comprises:
performing operations on the plurality of first coefficients according to the plurality of direction results, for respectively generating each of a plurality of first interpolated coefficients between the pixel locations corresponding to the plurality of first coefficients; and
generating the plurality of first destination coefficients with the plurality of first c

Assignees

• Novatek Microelectronics Corp

Inventors

• Zha Lin
• Chen Chun-Wei
• Du Yuanjia
• Wang Chun

Classifications

• G06F17/175 Primary

  of multidimensional data · CPC title

• G06T3/4007 Primary

  based on interpolation, e.g. bilinear interpolation (image demosaicing G06T3/4015; edge-driven or edge-based scaling G06T3/403) · CPC title

• G06T7/0026

  Physics · mapped topic

• G06K9/527

  Physics · mapped topic

• G06T2207/20064

  Wavelet transform [DWT] · CPC title