CN114372938A - Image self-adaptive restoration method based on calibration - Google Patents

Abstract

The invention belongs to the field of image restoration processing, and specifically discloses an image adaptive restoration method based on calibration, comprising the steps of: noise calibration, shooting and saving a set target, and performing noise calibration; imaging system calibration; image denoising processing; image deblurring; output the final restored image. The invention performs parameterized mathematical modeling by calibrating the noise and blur coefficient of the actual optical system, and applies the calibration result to the subsequent image restoration algorithm to obtain the best image restoration result for this imaging system, and greatly improves the image restoration. Effect.

Description

A Calibration-Based Adaptive Restoration Method for Images

technical field

The invention relates to the field of image restoration processing, in particular to an image adaptive restoration method based on calibration.

Background technique

The image formed by a point in object space through an ideal optical imaging system is still a point. However, the actual optical imaging system is affected by geometric aberration and optical diffraction limit, and the image is no longer clear and perfect, and there is a certain optical blur. At the same time, the imaging process is always accompanied by noise signals. The noise mainly comes from Poisson noise related to light intensity, transfer noise introduced in the charge transfer process, dark current noise generated by thermal excitation in the silicon substrate, and non-uniformity. noise, etc.

These factors lead to the degradation of image quality, so image denoising and deblurring are very important steps in image restoration. Image denoising generally starts from the smoothness of the image, the high frequency in the frequency domain and the randomness of the distribution, and constructs a corresponding mathematical model to remove noise; image denoising uses human cognition of clear images to add a priori to the degradation model Solve the model to achieve the function of image deblurring.

Most of the current algorithms artificially give certain assumptions to image processing, such as noise obeying Gaussian distribution or the default value of blur kernel, which will make the image restoration effect unsatisfactory for different imaging systems.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide an image adaptive restoration method based on calibration, so as to solve the problems raised in the above background art.

To achieve the above object, the present invention provides the following technical solutions: a calibration-based image adaptive restoration method, comprising the steps:

S1: Noise calibration, shoot and save the set target, and perform noise calibration;

S2: imaging system calibration;

S3: Image denoising processing;

S4: Image deblurring;

S5: Output the final restored image.

Preferably, the noise calibration in S1 includes global noise calibration and local noise calibration, and the specific steps of noise calibration include:

S101: Set the target, use the actual optical system to shoot the target, and save the captured image;

S102: Select two target images of the same scene at the same moment;

S103: Calculate the global image noise distribution;

S104: Calculate the local image noise distribution.

Preferably, the calculation process of the global image noise distribution in S103 specifically includes:

S103a: Perform a difference operation on the two frames of target images to obtain a difference image;

S103b: Perform distribution fitting of the probability density function on the difference image to determine its distribution type and specific parameter values;

S103c: save the result in the parameter file.

Preferably, the calculation process of the local image noise distribution in S104 specifically includes:

S104a: intercept N image blocks with different grayscale values on the target image, and N is 8 by default;

S104b: respectively draw a histogram for each image block;

S104c: Perform the distribution fitting of the probability density function again to determine its distribution type and specific parameter values;

S104d: average the results of N image blocks;

S104e: save all the results into the parameter file.

Preferably, the imaging system calibration process of S2 specifically includes:

S201: select a frame of target image;

S202: Capture a region ROI with black and white oblique edges in the image;

S203: "project" the data sequence of different rows in the ROI on the same pixel grid to obtain the edge spread function ESF;

S204: derive the edge spread function ESF to obtain the line spread function LSF of the rate of change of the straight line;

S205: Perform Fourier FFT transformation on the LSF to obtain the response value SFR at each spatial frequency;

S206: analyze the SFR data to obtain the blur parameters of the imaging system;

S207: Store the result in the parameter file.

Preferably, the image denoising process in S3 is implemented based on a distribution algorithm, and the distribution algorithm includes but is not limited to median filtering, non-local mean algorithm, and R-L algorithm.

Preferably, the image deblurring process in S4 is implemented based on deblurring calculation, and the image after deblurring is the final restored image.

Compared with the prior art, the beneficial effects of the present invention are:

The invention uses the current optical system to calibrate the noise distribution and the imaging system, saves the calibrated file, applies it to the subsequent algorithm as a priori parameter, and selects different algorithms according to the noise distribution, which can make the denoising effect the best; the imaging system is calibrated The result is used as the initial value of the edge deblurring algorithm, which further improves the accuracy of the blur kernel estimation for blind deblurring, which can greatly improve the deblurring effect.

Description of drawings

Fig. 1 is the overall flow chart of the present invention;

2 is a flowchart of noise calibration in an embodiment of the present invention;

3 is a flowchart of imaging system calibration in an embodiment of the present invention;

4 is a flowchart of image denoising in an embodiment of the present invention;

5 is a flowchart of image deblurring in an embodiment of the present invention;

FIG. 6 is a schematic diagram of a calibration target in an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Referring to FIG. 1, the present invention provides a technical solution: a calibration-based image adaptive restoration method, comprising the steps of:

S1: Noise calibration, shoot and save the set target, and perform noise calibration (the calibration target is shown in Figure 6);

S2: imaging system calibration;

S3: Image denoising processing;

S4: Image deblurring;

S5: Output the final restored image.

Please refer to FIG. 2. In this embodiment, the noise calibration in S1 includes global noise calibration and local noise calibration. The specific steps of noise calibration include:

S101: Set the target, use the actual optical system to shoot the target, and save the captured image;

S102: Select two target images of the same scene at the same moment;

S103: Calculate the global image noise distribution;

S104: Calculate the local image noise distribution.

In this embodiment, the calculation process of the global image noise distribution in S103 specifically includes:

S103a: Perform a difference operation on the two frames of target images to obtain a difference image;

S103b: Perform distribution fitting of the probability density function on the difference image to determine its distribution type and specific parameter values;

S103c: save the result in the parameter file.

In this embodiment, the calculation process of the local image noise distribution in S104 specifically includes:

S104a: intercept N image blocks with different grayscale values on the target image, and N is 8 by default;

S104b: respectively draw a histogram for each image block;

S104c: Perform the distribution fitting of the probability density function again to determine its distribution type and specific parameter values;

S104d: average the results of N image blocks;

S104e: save all the results into the parameter file.

Referring to FIG. 3, in this embodiment, the imaging system calibration process of S2 specifically includes:

S201: select a frame of target image;

S202: Capture a region ROI with black and white oblique edges in the image;

S203: "project" the data sequence of different rows in the ROI on the same pixel grid to obtain the edge spread function ESF;

S204: derive the edge spread function ESF to obtain the line spread function LSF of the rate of change of the straight line;

S205: Perform Fourier FFT transformation on the LSF to obtain the response value SFR at each spatial frequency;

S206: analyze the SFR data to obtain the imaging system blur parameters; S207: store the results in a parameter file.

Referring to FIG. 4 , in this embodiment, the image denoising process in S3 is implemented based on a distribution algorithm, and the distribution algorithm includes but is not limited to median filtering, non-local mean algorithm, and R-L algorithm.

S3 specifically includes steps:

S301: Import parameter file;

S302: select the input image to be processed;

S303: According to the noise distribution calibrated by the parameter file, select an adaptive algorithm, and the algorithm provides an adaptive algorithm selection; the algorithm includes:

a. Median filtering (salt and pepper noise), median filtering is a nonlinear signal processing technology based on sorting statistics theory that can effectively suppress noise. The basic principle of median filtering is to use the value of a point in a digital image or digital sequence with The median value of each point value in a neighborhood of the point is replaced, thereby eliminating isolated noise points;

b. Non-local mean algorithm (Gaussian noise), the non-local mean algorithm is a method of taking the mean value smoothly in the area around a target pixel, so the non-local mean filter means that it uses all the pixels in the image, these pixels are based on some similarity weighted average. After filtering, the image has high definition and no details are lost; the non-local mean algorithm uses redundant information ubiquitous in natural images to remove noise, which is different from bilinear filtering, median filtering, etc. The whole image is denoised, that is to find similar areas in the image in units of image blocks, and then average these areas to filter out the Gaussian noise in the image better;

c, R-L algorithm (Poisson noise);

S304: process the input image;

S305: Output the processed image, and then enter the image deblurring process.

Referring to FIG. 5 , in this embodiment, the image deblurring process in S4 is implemented based on deblurring calculation, and the image after deblurring is the final restored image.

In this embodiment, S4 specifically includes:

S401: Read the fuzzy parameters of the imaging system calibrated by the parameter file, and construct a two-dimensional kernel function;

S402: Use the kernel obtained in S401 as the initial value based on the edge deblurring algorithm, and start image deblurring;

S403: The restoration operation is completed after the image is de-blurred.

In the above embodiment, the parameterized mathematical modeling is performed by calibrating the noise and blur coefficient of the actual optical system, and the calibration result is applied to the subsequent image restoration algorithm to obtain the best image restoration result for this imaging system.

Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, and substitutions can be made in these embodiments without departing from the principle and spirit of the invention and modifications, the scope of the present invention is defined by the appended claims and their equivalents.

Claims (7)

1. an image adaptive restoration method based on calibration, is characterized in that, comprises the steps: S1: Noise calibration, shoot and save the set target, and perform noise calibration; S2: imaging system calibration; S3: Image denoising processing; S4: image deblurring; S5: Output the final restored image. 2. A calibration-based image adaptive restoration method according to claim 1, wherein the noise calibration in the S1 includes global noise calibration and local noise calibration, and the specific steps of noise calibration include: S101: Set the target, use the actual optical system to shoot the target, and save the captured image; S102: Select two target images of the same scene at the same moment; S103: Calculate the global image noise distribution; S104: Calculate the local image noise distribution. 3. a kind of image adaptive restoration method based on calibration according to claim 2, is characterized in that, in described S103, the global image noise distribution calculation process specifically comprises: S103a: Perform a difference operation on the two frames of target images to obtain a difference image; S103b: Perform distribution fitting of the probability density function on the difference image to determine its distribution type and specific parameter values; S103c: save the result in the parameter file. 4. a kind of image adaptive restoration method based on calibration according to claim 2, is characterized in that, in described S104, the local image noise distribution calculation process specifically comprises: S104a: intercept N image blocks with different grayscale values on the target image, and N is 8 by default; S104b: respectively draw a histogram for each image block; S104c: Perform the distribution fitting of the probability density function again to determine its distribution type and specific parameter values; S104d: average the results of N image blocks; S104e: save all the results into the parameter file. 5. A calibration-based image adaptive restoration method according to claim 1, wherein the imaging system calibration process of the S2 specifically comprises: S201: select a frame of target image; S202: Capture a region ROI with black and white oblique edges in the image; S203: "project" the data sequence of different rows in the ROI on the same pixel grid to obtain the edge spread function ESF; S204: derive the edge spread function ESF to obtain the line spread function LSF of the rate of change of the straight line; S205: Perform Fourier FFT transformation on the LSF to obtain the response value SFR at each spatial frequency; S206: analyze the SFR data to obtain the blur parameters of the imaging system; S207: Store the result in the parameter file. 6. a kind of image adaptive restoration method based on calibration according to claim 1, is characterized in that, in described S3, image denoising processing is realized based on distribution algorithm, and distribution algorithm includes but is not limited to median filter, non-local mean value Algorithm, R-L algorithm. 7. a kind of image adaptive restoration method based on calibration according to claim 1, is characterized in that, in described S4, image deblurring process is realized based on deblurring algorithm, and the picture after deblurring is final restoration post image.

Images