Automatic measurement method and system for the period and phase of an annular texture image

Abstract

The application discloses a kind of annular texture image period, phase automatic measurement method and system, the measurement method belongs to: S1. obtain annular texture image, and the inner and outer radius of annular texture area and the center are obtained after pre-processing;S2. polar coordinate transformation is carried out to annular region, and annular rectangular image is obtained by intercepting annular surface region;S3. process annular rectangular image signal, and obtain one-dimensional signal;S4. one-dimensional fourier transform is carried out to one-dimensional signal, and the frequency spectrum and phase spectrum of image are obtained;S5. the frequency spectrum rate of maximum frequency spectrum ability is taken as the period of texture image, and the phase of corresponding frequency in phase spectrum is calculated to obtain the phase of current texture period.The application is aimed at annular periodic texture data, polar coordinate transformation is used to obtain polar coordinate annular image, image signal is processed under polar coordinate, and the period and phase of image are measured quickly and accurately.

Description

Technical Field

[0001] This invention relates to the fields of computer vision inspection and image processing technology, and in particular to an automatic measurement method and system for the period and phase of a ring-shaped texture image. Background Technology

[0002] With the rapid development of the industrial manufacturing sector, the demand and application scenarios for industrial image defect detection have experienced explosive growth, and machine vision technology is increasingly being applied to various stages of industrial production. In the industrial field, annular periodic texture images frequently appear in production scenarios such as bearings, gaskets, and metal parts. Measuring the texture period and phase of such data is one of the crucial steps in defect detection. Traditional manual inspection methods suffer from high costs, low efficiency, and insufficient accuracy. With the rapid development of computer vision technology, a series of image period detection techniques and methods have emerged. However, previous methods have not solved the problem of period measurement for annular periodic texture data, nor have they determined the phase information of the texture data. Furthermore, they suffer from insufficient accuracy and limitations in the measurement period size, making it difficult to solve the current problem of period and phase detection for annular texture images.

[0003] Existing image period detection solutions address the problem of image period measurement in Cartesian coordinates. Chinese patent publication CN107037050A discloses an automatic method for measuring the period of fabric image texture. This method aims to solve the problem of measuring the row and column periods of knitted textures in Cartesian coordinates. It adjusts the two-dimensional periodic texture image into two sets of one-dimensional signals (rows and columns), performs Fast Fourier Transform on each set of signals to obtain the spectrum, and determines the period value in the row and column directions based on information such as the maximum spectral value. Chinese patent publication CN109781736A discloses an automatic method and system for measuring the period of wafer texture images. This method aims to measure the row and column periods of wafer images in Cartesian coordinates. The method performs Gaussian filtering on the wafer image, followed by a two-dimensional Fast Fourier Transform. Autocorrelation is calculated on the filtered image in the frequency domain, and the local maxima of the autocorrelation result are obtained. The distance between the maxima is the texture image period.

[0004] The drawback of the above techniques is that they cannot measure the period of a circular image and do not measure the initial phase information of the image.

[0005] Therefore, it is necessary to propose an automatic measurement method for the period and phase of a ring-shaped texture image. For ring-shaped periodic texture data, polar coordinate transformation is used to obtain a polar coordinate ring image, and the image signal is processed in polar coordinates to quickly and accurately measure the period and phase of the image. Summary of the Invention

[0006] The purpose of this invention is to provide an automatic measurement method and system for the period and phase of a ring-shaped texture image, so as to solve one of the many problems described above.

[0007] In view of this, the solution of the present invention is as follows:

[0008] An automatic measurement method for the period and phase of a ring-shaped texture image includes the following steps:

[0009] S1. Obtain the ring-shaped texture image, and preprocess it to obtain the inner and outer radii and center of the ring-shaped texture region;

[0010] S2. Perform polar coordinate transformation on the annular region and extract the annular region to obtain a rectangular image of the annular surface;

[0011] S3. Process the toroidal rectangular image signal to obtain a one-dimensional signal;

[0012] S4. Perform a one-dimensional Fourier transform on the one-dimensional signal to obtain the spectrum and phase spectrum of the image;

[0013] S5. The spectral rate with the greatest spectral capability is the period of the texture image. The phase of the current texture period is calculated based on the phase of the corresponding frequency in the phase spectrum.

[0014] Further, step S3 calculates the mean of each row of data in the toroidal rectangular image to obtain a one-dimensional signal.

[0015] Preferably, step S3 first applies Gaussian filtering to the toroidal rectangular image to remove noise.

[0016] Preferably, step S3 involves removing the mean from the one-dimensional signal.

[0017] Further, in step S5, the spectrum of the image obtained by the one-dimensional Fourier transform is obtained by calculating the energy calculation function, and the phase spectrum is obtained by calculating the phase calculation function.

[0018] Further, in step S5, the phase of the texture image under the period is divided by the period to obtain the phase of the current texture period.

[0019] The present invention also proposes a measurement system for the above-described automatic measurement method, comprising:

[0020] The ring extraction module is used to obtain ring-shaped texture images;

[0021] The polar coordinate transformation module performs polar coordinate transformation on the annular region and obtains a toroidal rectangular image.

[0022] The signal processing module processes the toroidal rectangular image signal to obtain a one-dimensional signal;

[0023] The Fourier transform module performs a one-dimensional Fourier transform on a one-dimensional signal to obtain the spectrum and phase spectrum of the image.

[0024] The calculation module is used to calculate the period of the texture image and the phase of the current texture period.

[0025] Furthermore, the signal processing module includes a noise reduction module, a one-dimensional processing module, and a mean removal module; the noise reduction module performs Gaussian filtering on the toroidal rectangular image to remove noise, the one-dimensional processing module calculates the mean of each row of data in the toroidal rectangular image to obtain a one-dimensional signal, and the mean removal module performs mean removal processing on the one-dimensional signal.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] 1. This invention solves the problem of automatic periodic detection in circular periodic texture images and proposes a fast and accurate detection scheme;

[0028] 2. This invention detects the initial phase information of the annular texture while simultaneously detecting the detection cycle;

[0029] 3. The method proposed in this invention has error correction capabilities and can accurately measure the period and phase even for defective data. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is an overall structural diagram of the automatic measurement system described in this invention.

[0032] Figure 2 This is a structural diagram of the signal processing module described in this invention.

[0033] Figure 3 This is an overall flowchart of the automatic measurement method described in this invention.

[0034] Figure 4 These are schematic diagrams of normal and defective circular texture images of the present invention.

[0035] Figure 5 This is the spectral result obtained by processing normal and defective circular texture images according to the present invention.

[0036] Figure 6 This is a schematic diagram of the texture period and phase effect calculated based on the spectrum results of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described in this specification are merely for explaining the invention and are not intended to limit the invention.

[0038] like Figure 1 As shown, this invention proposes an automatic measurement system for the period and phase of a ring-shaped texture image, comprising:

[0039] The annular extraction module 100 is used to acquire an annular texture image;

[0040] Polar coordinate transformation module 200 performs polar coordinate transformation on the annular region and obtains a toroidal rectangular image;

[0041] Signal processing module 300 processes the toroidal rectangular image signal to obtain a one-dimensional signal;

[0042] The Fourier transform module 400 performs a one-dimensional Fourier transform on a one-dimensional signal to obtain the spectrum and phase spectrum of the image.

[0043] The calculation module 500 is used to calculate the period of the texture image and the phase of the current texture period, including calculating the frequency domain energy spectrum and phase spectrum by using the energy calculation function and the phase calculation function respectively.

[0044] like Figure 2 As shown, the signal processing module 300 includes a noise reduction module 301, a one-dimensional processing module 302, and a mean removal module 303. The noise reduction module 301 performs Gaussian filtering on the toroidal rectangular image to remove noise. The one-dimensional processing module 302 calculates the mean of each row of data in the toroidal rectangular image to obtain a one-dimensional signal. The mean removal module 303 performs mean removal processing on the one-dimensional signal to avoid the original data having too much low-frequency energy affecting periodic analysis.

[0045] This invention proposes an automatic measurement method for the period and phase of a ring-shaped texture image, the flowchart of which is shown below. Figure 3 As shown, the method includes the following steps:

[0046] S1. Obtain the ring-shaped texture image and preprocess the image data I to obtain the inner and outer radii of the ring-shaped texture region. r 1 , r 2 and the center c(x, y) The methods include, but are not limited to, Hough transform, image processing, and other methods;

[0047] S2. Perform polar coordinate transformation on the annular region data, and then crop the annular region according to the relationship between the inner and outer radii of the annulus to obtain a rectangular image of the annulus:

[0048] a) After transformation, the result is d×d The polar coordinate image P can be transformed into rectangular data by polar coordinate transformation, which facilitates feature analysis.

[0049] b) Extract from The beginning Image of size P c Only the core data of the circular part is retained;

[0050] S3. Processing toroidal rectangular images P c One-dimensional signal is obtained. M´ :

[0051] a) to P c Perform Gaussian filtering to remove noise and obtain P c ´ ;

[0052] b) To avoid the excessive low-frequency energy in the original data affecting periodic analysis, the original data needs to be mean-reduced to obtain a signal with a mean of 0. Calculation P c ´ The mean of each row of data yields one-dimensional data. M ;

[0053]

[0054] c) Data M After performing the "mean removal" process, the following is obtained: M´ ;

[0055]

[0056] S4. For the processed one-dimensional signal M´ Performing a one-dimensional Fourier transform will significantly reduce the influence of low-frequency energy, thus obtaining the image's spectrum. E and phase spectrum G :

[0057] a) One-dimensional Fourier transform, yielding d 3D complex vector X :

[0058]

[0059] in j Represents the imaginary unit.d For signal length, e Let be the natural logarithm constant, for each distinct k The values ​​are calculated, and the results are summarized to obtain a complex vector. X .

[0060] b) Based on complex vectors X The frequency domain energy spectrum is obtained by combining the energy calculation function. E ;

[0061]

[0062] abs This represents the energy calculation function.

[0063] c) Based on complex vectors X The phase spectrum is obtained by combining the phase calculation function. G :

[0064]

[0065] angle This represents the phase calculation function.

[0066] S5. Take the frequency with the highest spectral energy as the period of the texture image. c :

[0067]

[0068] S6. Perform corresponding processing on the phases of the corresponding frequencies in the phase spectrum, and use the result as the phase of the current texture period:

[0069]

[0070] In one specific embodiment, such as Figure 2-4 As shown, for a normal annular texture image (4A) and an image with severely missing data (4B) of the same workpiece, the spectrum of the obtained images obtained by using the automatic measurement method for the period and phase of the annular texture image described in this invention corresponds to the following: Figure 5 (A) Figure 5 (B); The example workpiece cycle analysis effect diagrams are respectively as follows: Figure 6 (A) Figure 6 (B). From Figure 5 (B) It is easy to see that even for images with severely missing data, the period corresponding to the maximum signal strength is still 16, and the measurement method provided by this invention can still restore the correct period. Furthermore, combining period and phase, an example workpiece period analysis effect diagram is given ( Figure 6 As can be seen from the image, the initial phase information of the ring texture can be obtained accordingly, and the image with severely missing data is completely consistent with the information of the normal ring texture image.

[0071] Although embodiments of the invention have been shown and described, other advantages and modifications will be readily apparent to those skilled in the art. Therefore, the invention is not limited to the specific details, representative devices and illustrated examples shown and described herein without departing from the spirit and scope of the general concept as defined by the claims and their equivalents.

Claims

1. A method for automatic measurement of the period and phase of an annular texture image, characterized in that, The method comprises the following steps: S1. Obtain a ring-shaped texture image, and pre-process to obtain the inner and outer radii and the center of the ring-shaped texture region; S2. Perform polar coordinate transformation on the ring-shaped region, and obtain a ring surface rectangular image by intercepting the ring surface region; S3. Calculate the mean value of each row of data of the ring surface rectangular image to obtain a one-dimensional signal; S4. Perform one-dimensional Fourier transformation on the one-dimensional signal to obtain the frequency spectrum and the phase spectrum of the image; S5. Take the frequency spectrum with the maximum energy as the period of the texture image, and take the phase corresponding to the frequency in the phase spectrum as the phase of the current texture period.

2. The automatic measuring method according to claim 1, characterized in that, The step S3 performs Gaussian filtering on the ring surface rectangular image to remove noise.

3. The automatic measurement method according to claim 1, characterized in that, The step S3 performs mean value processing on the one-dimensional signal.

4. The automatic measurement method according to claim 1, characterized in that, The step S4 calculates the frequency spectrum according to an energy calculation function and calculates the phase spectrum according to a phase calculation function.

5. A measurement system performing the method of claim 1, characterized by The method comprises: a ring extraction module (100) configured to obtain a ring-shaped texture image; a polar coordinate transformation module (200) configured to perform polar coordinate transformation on the ring-shaped region and obtain a ring surface rectangular image; a signal processing module (300) configured to process the signal of the ring surface rectangular image to obtain a one-dimensional signal; a Fourier transformation module (400) configured to perform one-dimensional Fourier transformation on the one-dimensional signal to obtain the frequency spectrum and the phase spectrum of the image; a calculation module (500) configured to calculate the period of the texture image and the phase of the current texture period.

6. The measurement system of claim 5, wherein, The signal processing module (300) comprises a noise removal module (301), a one-dimensional processing module (302), and a mean value removal module (303). The noise removal module (301) performs Gaussian filtering on the ring surface rectangular image to remove noise. The one-dimensional processing module (302) calculates the mean value of each row of data of the ring surface rectangular image to obtain a one-dimensional signal. The mean value removal module (303) performs mean value processing on the one-dimensional signal.

Images