Auto-collimator noise error compensation method based on Retinex theory

Abstract

The invention discloses an autocollimator noise error compensation method based on a Retinex theory, and belongs to the field of optical measuring instruments. The method comprises the following steps of: 1, combining a Retinex theoretical model with an auto-collimation measurement system, and correspondingly analyzing light source emergent light, system measurement light beams and background noise light of an auto-collimator and an input image, a reflection image and a brightness image of the Retinex theory from the angle of identifying a detection image by a photoelectric detector in an auto-collimation system structure; 2, performing gray processing and logarithmic processing on the acquired input image and the brightness image, and obtaining a difference value to obtain a reflection gray image with image information of the object; and 3, a contrast test is designed for the input image and the reflection image before and after the noise suppression scheme is adopted, the displacement of the diaphragm image in the two images is detected and identified, the error compensation method is analyzed to effectively suppress the interference of background noise on imaging, and the measurement precision and the measurement stability of the autocollimator are improved.

Description

Technical Field

[0001] The present invention belongs to the field of optical measuring instruments, and specifically proposes an image enhancement algorithm model based on the Retinex theory to enhance the anti-interference ability of the autocollimator to noise. Background Technique

[0002] The autocollimation technology is the main method for realizing high-precision micro-angle measurement, and can complete the conversion from the angle change amount to the linear measurement. The ELCOMAT3000 model autocollimator produced by the German company Moeller-Wedel has achieved a resolution of 0.005″, and the measurement accuracy is within ±0.25″ within the range of ±1000″. Now, the photoelectric sensor has reached the sub-pixel level of spot displacement detection ability, making the photoelectric autocollimator the angle measurement device with the highest accuracy and resolution at present, and occupying an important position in the field of precision measurement.

[0003] In response to the requirements for higher precision and anti-interference performance, a large number of studies on the accuracy improvement, error analysis and reduction of the autocollimation measurement system have been carried out by multiple research institutes worldwide.

[0004] Zheng (Literature: Ligong Z, Hai Z, Erhui Q, et al. Characterized environmental influences on autocollimator measurement uncertainty using an extended Allan variance[J]. Optics and Lasers in Engineering, 2024, 172) et al. proposed an extended Allan variance based on a point-by-point scanning measurement mode to characterize the environmental influence on AC measurement in view of the influence of time-varying errors on the accuracy and reading stability of the autocollimator. By optimizing the measurement strategy, the measurement uncertainty of the autocollimator was improved to 2.3 nrad RMS. Li (Literature: Zhongtang L, Mingyao L, Jianfeng L. High Precision Autocollimation Measurement Technology Based on Image Recognition[C] / / Eliwise Academy. Proceedings of the 2nd International Conference on Computing and Data Science (CONF-CDS2021). Shanghai Normal University TIANHUA College; Wuhan University of Technology; Tongji University;, 2021:5.) et al. separated the cross target image based on a dynamic threshold and used an image recognition algorithm to assist in calibrating the measurement target of the autocollimator, improving the accuracy of the optical axis eccentricity error to 0.1 pixel.

[0005] In previous studies, most improvements and optimizations of measurement accuracy were based on the imaging aberration of the autocollimator optical system. The influence of the autocollimator's own structural problems on measurement accuracy was rarely explored, and the reflection noise caused by the non-working structural surfaces of the optical elements in the autocollimator was ignored. With the development of sensor pixel subdivision technology, the influence of this noise error source on autocollimation measurement accuracy will be gradually amplified, seriously affecting the discrimination of the detection image by the optoelectronic sensor.

[0006] In the present invention, the influence of reflection noise on the imaging of the measurement beam is analyzed according to the Retinex theory, and a denoising algorithm model based on the Retinex theory is disclosed for such an error source. A comparative experiment is carried out on the autocollimation system before and after the measurement error is compensated by the algorithm. The experimental results show that this error source has a significant impact on the measurement accuracy of the autocollimator, and the proposed error compensation scheme has an obvious effect on improving the accuracy. At the same time, the stability of the autocollimation system for single data acquisition is also improved. Summary of the Invention

[0007] The present invention aims to solve the problem that the background noise affects the measurement accuracy of the autocollimator due to its own structure. According to the Retinex theory model, the optical signals in different stages of the autocollimator measurement model are classified and analyzed to enhance the recognition difference between the photodetector for background noise and the effective optical signal, so as to more effectively weaken or suppress the background noise and improve the accuracy of the correct image. The technical solution of the present invention is as follows:

[0008] An autocollimator noise error compensation method based on the Retinex theory, characterized by comprising the following steps:

[0009] Step 1: Construct a measurement model adapted to the Retinex theory in the autocollimation system;

[0010] Step 2: Classify and analyze the light beams in each stage of the autocollimator, and correspondingly combine them with the three components of the luminance image, reflection image, and input image in the Retinex theory model;

[0011] Step 3: Collect the luminance image and the input image, and perform gray value processing on the images;

[0012] Step 4: Perform difference processing on the two gray images through the Matlab algorithm to obtain the reflection image;

[0013] Step 5: Use the Hough transform to detect the cross imaging spot in the reflection image to obtain the intersection coordinates

[0014] Step 6: Set up a control experimental group, one group is an autocollimation measurement system with enhanced image recognition, and the other group is an original autocollimation measurement system. Rotate the reflector of the autocollimation system multiple times, record the intersection coordinates, and analyze the effect of this method on suppressing the internal reflection noise of the autocollimator for the data of the two groups.

[0015] Furthermore, mainly aiming at the phenomenon of stray light generated by the reflection of other optical elements such as beam splitters, the non-working optical structure surfaces result in multiple noise optical paths, interfering with the imaging effect of the normal measurement beam. Among them, the lower surface, the rear surface of the beam splitter and the front surface of the collimator have relatively more reflection effects. Regarding the problem that the reflection phenomenon of other optical elements such as beam splitters affects the measurement accuracy of the autocollimator, specifically, the background noise is unevenly distributed on the image plane, resulting in errors in the detector's recognition of the diaphragm image.

[0016] Selecting a diaphragm in the shape of a cross can enable the detector to more accurately perceive the displacement of the diaphragm pattern on the imaging plane.

[0017] Furthermore, the above-mentioned background noise will cause the detector to deviate from the ideal straight line when selecting the horizontal points (x 1 , y 1 ) and (x 2 , y 2 ) of the cross, and the vertical points (x 3 , y 3 ) and (x 4 , y 4 ). Then, an error straight line is determined through these two points.

[0018] The detector identifies the detection points through the Hough transform, then converts from the (ρ, θ) domain to the rectangular coordinate system and completes the confirmation of the straight line. The straight line equation of the horizontal line obtained by the method of undetermined coefficients can be expressed as:

[0019]

[0020] Similarly, the straight line equation of the vertical line can be expressed as:

[0021]

[0022] By combining Equation (1) and Equation (2), the cross intersection coordinates (X I , Y I ) with errors can be obtained:

[0023]

[0024] Furthermore, the Retinex theory believes that an image can be decomposed into two parts: the luminance image L(x, y) containing the incident light information, which includes the ambient illumination conditions; and the reflection image R(x, y) containing the object's own image information. And the input image I(x, y) received by the detector is the convolution of the two, which can be expressed as:

[0025] I(x, y) = R(x, y) × L(x, y) (3)

[0026] However, when humans observe, the recognition of images is usually in a superposition state of multiple images. Therefore, it is easier to understand by taking the logarithmic processing of formula (3), that is:

[0027] ln[I(x,y)] = ln[R(x,y)] + ln[L(x,y)] (4)

[0028] Furthermore, when combining the autocollimation measurement model and the Retinex theory model, it is necessary to perform corresponding analysis on the light emitted by the light source of the autocollimator, the system measurement beam, the background noise light, and the input image, reflection image, and brightness image of the Retinex theory. It is known that the incident light passes through the aperture stop multiple times for transmission and reflection, and finally focuses on the detector for imaging. This part of the beam reflects the shape attributes of the aperture pattern of the measurement target and can be regarded as the reflection image R(x,y) in the Retinex model. Other noise beams form a large-range interference image on the CCD image plane. These beams more reflect the illumination conditions when the detector acquires the image and can be regarded as the brightness image L(x,y). The mixed image detected on the final imaging plane is the input image I(x,y).

[0029] Perform gray-scale processing on the available brightness image and input image, divide the detected light intensity according to the pixel threshold set by the algorithm, from 0 to 2 n Traverse the pixels at each region position in the gray-scale image to complete the gray-scale processing of the image.

[0030] After completing the gray-scale processing of the brightness image and the input image, take the logarithm of the gray-scale values in the obtained gray-scale image and take the difference between the two. The process can be expressed as:

[0031] ln[I(x,y)] - ln[L(x,y)] = ln[R(x,y)] (5)

[0032] After obtaining the gray-scale image of the reflection image, then identify the detection pattern therein. Identify the intersection coordinates of the reflection image, and then take the difference with the intersection coordinates with errors, and the effect of this image enhancement type of noise compensation scheme on suppressing noise interference and improving measurement accuracy can be obtained.

[0033] An electronic device, characterized in that it includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the autocollimator noise error compensation method based on the Retinex theory according to any one of claims 1 to 4.

[0034] A non-transitory computer-readable storage medium, on which a computer program is stored, characterized in that when the computer program is executed by a processor, it implements the method for compensating the noise error of an autocollimator based on the Retinex theory as described in any one of claims 1 to 6.

[0035] A computer program product, including a computer program, characterized in that when the computer program is executed by a processor, it implements the method for compensating the noise error of an autocollimator based on the Retinex theory as described in any one of claims 1 to 6.

[0036] The advantages and beneficial effects of the present invention are as follows:

[0037] The innovation of the present invention is specifically to combine the Retinex theory model with the autocollimation measurement system. Starting from the perspective of the photodetector's recognition of the detection image in the autocollimation system structure, the light emitted by the light source of the autocollimator, the system measurement beam, the background noise light, and the input image, reflection image, and luminance image of the Retinex theory are analyzed correspondingly. By enhancing the difference between the aperture detection pattern and the background noise, the background noise is better eliminated, and the interference in the photoelectric detection plane is weakened. This noise suppression scheme has a simple application form and has an obvious effect on improving the measurement accuracy of the autocollimator. Description of the Drawings

[0038] Figure 1 is the ideal optical path structure of the autocollimation system according to the embodiment of the present invention;

[0039] Figure 2 is a schematic diagram of the influence of background noise on image recognition according to the embodiment of the present invention;

[0040] Figure 3 is a schematic diagram of the application of the Retinex theory model in the autocollimation measurement system according to the embodiment of the present invention;

[0041] Figure 4 is a flowchart of image enhancement processing according to the embodiment of the present invention; Detailed Embodiments

[0042] Next, the technical solutions in the embodiments of the present invention will be clearly and detailedly described in conjunction with the drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention.

[0043] The technical solution for the present invention to solve the above technical problems is:

[0044] A method for compensating the noise error of an autocollimator based on the Retinex theory, including the following steps:

[0045] Step 1: Construct a measurement model adapted to the Retinex theory in the autocollimation system;

[0046] Step 2: Classify and analyze the light beams at each stage in the autocollimator, and correspondingly combine them with the three components of the luminance image, reflection image, and input image in the Retinex theory model;

[0047] Step 3: Collect the luminance image and the input image, and perform gray value processing on the images;

[0048] Step 4: Perform difference processing on the two gray images through the Matlab algorithm to obtain the reflection image;

[0049] Step 5: Use the Hough transform to detect the cross imaging spot in the reflection image to obtain the intersection coordinates

[0050] Step 6: Set up a control experimental group, one group is the autocollimation measurement system with image recognition enhancement, and the other group is the original autocollimation measurement system. Rotate the reflector of the autocollimation system multiple times, record the intersection coordinates, and analyze the data of the two groups to analyze the effect of this method on suppressing the internal reflection noise of the autocollimator.

[0051] Ideally, the imaging optical path of the classical autocollimator is as shown in the appendix Figure 1 There is no optical reflection other than the transmission of the working structure surface. However, the reflection of other optical elements such as the beam splitter generates stray light, resulting in multiple noise optical paths, interfering with the imaging effect of the normal measurement beam. Among them, the reflections with a relatively large influence are the lower surface, the rear surface of the beam splitter, and the front surface of the collimator. For the problem that the reflection of other optical elements such as the beam splitter affects the measurement accuracy of the autocollimator, specifically, the background noise is unevenly distributed on the image plane, resulting in errors in the detector's recognition of the aperture image.

[0052] Selecting a diaphragm in the shape of a cross can enable the detector to more accurately perceive the displacement of the diaphragm pattern on the imaging plane.

[0053] The above-mentioned background noise will cause the detector to deviate from the edge line where the energy is concentrated when selecting the horizontal and vertical points of the cross, and deviate outwards. As shown in the appendix Figure 2 As shown, the four selected points (x 1 , y 1 ), (x 2 , y 2 ), (x 3 , y 3 ), (x 4 , y 4 ) deviate from the ideal straight line, and two error straight lines are determined through these four points.

[0054] The detector uses the Hough transform to identify the detection points, then converts from the (ρ, θ) domain to the rectangular coordinate system and completes the confirmation of the straight line. The straight line equation of the horizontal line obtained by the method of undetermined coefficients can be expressed as:

[0055]

[0056] Similarly, the straight line equation of the vertical line can be expressed as:

[0057]

[0058] By combining equations (1) and (2), we can find the coordinates of the cross intersection with error (X I ,Y I ):

[0059]

[0060] Retinex theory believes that an image can be decomposed into two parts: the brightness image L(x,y) containing the incident light information, including the ambient lighting conditions; and the reflection image R(x,y) containing the image information of the object itself. The input image I(x,y) received by the detector is the convolution of the two, which can be expressed as:

[0061] I(x,y)=R(x,y)×L(x,y) (3)

[0062] However, when humans observe an image, they usually recognize the superposition of multiple images. Therefore, it is easier to understand if we perform logarithmic processing on formula (3), that is:

[0063] ln[I(x,y)]=ln[R(x,y)]+ln[L(x,y)] (4)

[0064] As attached Figure 3 As shown in the figure, when the autocollimation measurement model is combined with the Retinex theoretical model, the output light of the autocollimator, the system measurement beam and the background noise light need to be analyzed in correspondence with the input image, reflection image and brightness image of the Retinex theory. It is known that the incident light passes through the aperture diaphragm for multiple transmissions and reflections, and is finally focused on the detector to form an image. This part of the light beam reflects the shape properties of the aperture pattern of the measurement target, and can be regarded as the reflection image R(x,y) in the Retinex model. The other noise beams form a large-scale interference image on the CCD image plane. These beams more reflect the lighting conditions of the detector when acquiring the image, and can be regarded as the brightness image L(x,y). The mixed image detected on the final imaging plane is the input image I(x,y).

[0065] The available brightness image and input image are gray-scaled, and the detected light intensity is divided according to the pixel threshold set by the algorithm, from 0 to 2 n Traverse the pixels in each area of ​​the grayscale image to complete the grayscale processing of the image.

[0066] After the grayscale processing of the luminance image and the input image is completed, the subsequent process is as shown in the appendix Figure 4 As shown, take the logarithm of the grayscale values in the obtained grayscale image and take the difference between the two. The process can be expressed as:

[0067] ln[I(x,y)] - ln[L(x,y)] = ln[R(x,y)] (5)

[0068] After obtaining the grayscale image of the reflection image, then identify the detection pattern therein. Identify the intersection coordinates of the reflection image, and then take the difference with the intersection coordinates with errors, and the effect of this image enhancement type of noise compensation scheme on suppressing noise interference and improving measurement accuracy can be initially obtained.

[0069] Preferably, when experimentally verifying the above noise suppression scheme, in order to obtain the luminance image and the input image in the Retinex theoretical model, corresponding experimental equipment needs to be built on the optical platform. As shown in the appendix Figure 1 As shown, a photodetector, a beam splitter, a collimating objective lens, and a reflector are arranged in this order on the main optical axis, and a light source and a graticule are arranged at the orthogonal position of the beam splitter.

[0070] Preferably, a self-collimator with higher precision than the self-built self-collimator is required as a reference instrument to calibrate the measurement accuracy. At the same time, the reflector should be a cube reflector, and at least two adjacent sides are coated with reflective films. One side is aligned with the beam emitted from the inside of the self-collimation system, and the other side is aligned with the exit surface of the reference instrument. When rotating the reflector, record the angular displacement readings of the self-built self-collimator and the reference instrument at the same time.

[0071] Preferably, to obtain the luminance image, the normal imaging optical path needs to be blocked to ensure that only background noise is imaged on the detection plane, while to obtain the input image, the imaging of the mixed signal on the photoelectric detection plane is obtained normally.

[0072] It is proved by experiments that this compensation algorithm can effectively suppress the influence of noise on imaging detection. The measurement accuracies of the self-collimator around the X-axis and Y-axis can be improved from 4.29″ and 3.87″ to 3.59″ and 3.15″, and the stability performance of the system measurement is also improved. The stability performances in the X-axis and Y-axis directions are respectively improved by about 11.99% and 15.75%.

[0073] The installation of the control terminal, detection device, and optical device illustrated in the above embodiments, as well as their experimental detection, can specifically be realized by connecting a computer and the physical devices used in the experiment to achieve control and operation, or can be realized by a product with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0074] Computer-readable media includes both permanent and non-permanent, removable and non-removable media and can store information by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette tapes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transitory media that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0075] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or also includes elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising the said element.

[0076] The above embodiments should be understood as being only for illustrative purposes of the present invention and not for limiting the protection scope of the present invention. After reading the content described in the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A method for compensating autocollimator noise error based on Retinex theory, characterized in that: The following steps are involved: Step 1: Construct a measurement model adapted to the Retinex theory in the autocollimation system; Step 2: Classify and analyze the beams at each stage in the autocollimator, and combine them with the three components of the brightness image, reflection image and input image in the Retinex theoretical model; Step 3: Collect the brightness image and input image, and process the image into grayscale values; Step 4: Perform difference processing on the two grayscale images through Matlab algorithm to obtain the reflection image; Step 5: Use Hough transform to detect the cross imaging spot in the reflected image and obtain the intersection coordinates Step 6: Set up a control experimental group, one group is the autocollimation measurement system enhanced by image recognition, and the other group is the original autocollimation measurement system. Rotate the reflector of the autocollimation system multiple times, record the coordinates of the intersection, and combine the data of the two groups to analyze the effect of this method on suppressing the internal reflection noise of the autocollimator.

2. The autocollimator error according to claim 1 is mainly aimed at the phenomenon of stray light generated by reflection from other optical components such as a beam splitter, and the non-working optical structure surface causes multiple noise light paths to appear, interfering with the imaging effect of the normal measurement beam.

3. According to the autocollimator error described in claim 1, the problem of the reflection phenomenon of other optical components such as a beam splitter affecting the measurement accuracy of the autocollimator is specifically manifested in that the background noise is unevenly distributed on the image plane, resulting in errors in the detector's recognition of the straight line. The horizontal points (x1, y1), (x2, y2) and vertical points (x3, y3), (x4, y4) of the cross will deviate from the ideal straight line, and then an error line is determined through these two points.

4. The error compensation method based on Retinex theory according to claim 1, characterized in that: Retinex theory believes that an image can be decomposed into two parts: the brightness image L(x,y) containing the incident light information, reflecting the ambient lighting conditions when the image was acquired, and the reflection image R(x,y) containing the image information of the object itself. The input image I(x,y) received by the detector is the convolution of the two.

5. The autocollimator noise error compensation method based on Retinex theory according to claim 1, characterized in that: In step 2, the autocollimation measurement model is combined with the Retinex theoretical model. It is necessary to analyze the correspondence between the output light of the autocollimator, the system measurement beam and the background noise light and the input image, reflection image and brightness image of the Retinex theory.

6. The method for compensating the noise error of an autocollimator based on the Retinex theory according to claim 1, characterized in that, in step 3, the image is gray-scaled, and the detected light intensity is divided according to the pixel threshold value set by the algorithm, from 0 to 2. n Traverse the pixels in each area of ​​the grayscale image to complete the grayscale processing of the image.

7. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the autocollimator noise error compensation method based on the Retinex theory as claimed in any one of claims 1 to 6 is implemented.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the autocollimator noise error compensation method based on Retinex theory as claimed in any one of claims 1 to 6 is implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the autocollimator noise error compensation method based on Retinex theory as claimed in any one of claims 1 to 6 is implemented.

Images