RGBW mage demosaicing and deblurring combined method

A de-mosaic and de-blur technology, applied in image enhancement, image data processing, instruments, etc., can solve the problems of ringing phenomenon, inaccurate image blur kernel estimation, affecting the quality of image de-blurring, etc. the effect of clarity

Inactive Publication Date: 2018-01-23
成都精工华耀科技有限公司
View PDF2 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in the demosaic process of RGBW image to RGB image conversion, it is easy to introduce image noise between different channels, making the subsequent image blurring kernel estimation inaccurate, thus affecting the image deblurring quality, such as ringing phenomenon

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • RGBW mage demosaicing and deblurring combined method
  • RGBW mage demosaicing and deblurring combined method
  • RGBW mage demosaicing and deblurring combined method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0027] S1: According to the RGBW imaging format, extract y RGBW The r, g, b, and w channel pixels in the image constitute 4 single-channel images: y R 、y G 、y B 、y W , where y R 、y G 、y B 、y W The image size is y RGBW 1 / 4 of

[0028] S2: Estimate image y using blind convolution estimation method R 、y G 、y B 、y W The blur kernel: k r 、k g 、k b 、k w , such that:

[0029]

[0030]

[0031]

[0032]

[0033] where x r 、x g 、x b 、x w are respectively y R 、y G 、y B 、y W Corresponding sharp image, n r , n g , n b , n w is the image noise, the size of the blur kernel is: R*R;

[0034] S3: Create a matrix M of 2R*2R size, according to the RGBW imaging format, k r 、k g 、k b 、k w The middle element looks at the r, g, b, and w pixel values ​​and fills them in the matrix M in turn, making M a RGBW format fuzzy kernel, that is, the 2*2 elements in M ​​contain the r, g, b, and w channels fuzzy kernel value;

[0035] S4: Look at the matrix M in t...

Embodiment 2

[0039] The difference from Example 1 is that in step S5, the r, g, and b channel data in M' are taken as the blur kernels of the r, g, and b channel pixels respectively, and the image y RGB The r, g, and b channels are deconvoluted to obtain a clear image x RGB .

Embodiment 3

[0041] In step S1, the blind convolution proposed by Krishnan is used to estimate the blur kernel corresponding to the image, and the objective function of blind convolution to solve the blur kernel problem can be expressed as:

[0042]

[0043] Among them, k represents the blur kernel, also known as the point spread function PSF; x represents a clear image; y Indicates the blurred image obtained in S1; Represents a convolution operation; the first term Is the data fitting item, indicating the degree of matching between the blurred image and the clear image after convolution; the second item It is the quotient of the one-norm and two-norm of x, and it is the constraint item for the clear image x; the third item μ||k|| 1 Represents the prior knowledge of the total variant of the fuzzy kernel k, which is a constraint item for the fuzzy kernel k; λ and μ are the weight coefficients of the data fitting item and the constraint item; are energy conservation and non-negativ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

For the problems that, for an existing method, which is characterized by carrying out demosaicing processing on an RGBW mage first, and then, carrying out image deblurring processing, PSF estimation accuracy of is poor and it is easy to produce ringing phenomenon, the invention provides a method for carrying out demosaicing and deblurring simultaneously. The method comprises the following specificsteps: estimating an image fuzzy kernel on an RGBW single-channel image; carrying out demosaicing interpolation on the image fuzzy kernel; carrying out demosaicing on the RGBW image; and carrying outdeblurring on the image obtained after demosaicing to obtain a clear image. Compared with the prior art, the method can obviously improve PSF estimation accuracy and improve color imaging quality ofan RGBW image sensor under a simple lens system.

Description

technical field [0001] The invention relates to the technical field of digital image processing, in particular to a combined demosaicing and deblurring method for RGBW images. Background technique [0002] In modern optical systems, image quality will be reduced due to optical errors, and most single-convex lenses with spherical mirror structures will be affected by such as chromatic aberration, spherical aberration, and coma. In order to solve this dilemma, the existing optical imaging system mainly compensates for the aberration of the single lens through complex combination lenses. For example, the lens of a SLR camera may contain dozens of independent single lenses or lens groups. However, while the design of the complex combination lens improves the imaging quality, it undoubtedly greatly increases the cost of lens design and manufacture, and the volume and weight of the lens also increase thereupon. Therefore, how to reduce the design and manufacturing cost of the len...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06T5/00
Inventor 不公告发明人
Owner 成都精工华耀科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products