Method and system for measuring far-focus blurring kernel of optical imaging system
By using collimators and star plates to simulate infinity illumination, combined with the refinement method and standardization of the central sub-connection region, the problem of accurate measurement of the far-focus blur kernel in optical imaging systems was solved, improving image restoration effect and real-time performance.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HANGZHOU DIANZI UNIV
- Filing Date
- 2025-12-23
- Publication Date
- 2026-04-21
AI Technical Summary
Existing technologies struggle to accurately estimate the blur kernel of optical imaging systems under telephoto conditions, which affects image restoration performance.
By using a collimator and a star plate to simulate starlight at infinity, blur kernel images of various regions of the lens are acquired. The blur kernel is then accurately measured through a blur kernel refinement method and standardization processing of the central sub-connected region.
It improves the accuracy of fuzzy kernel measurement and image restoration performance, and realizes fast, real-time fuzzy kernel acquisition, which is suitable for real-time image processing.
Smart Images
Figure CN121898751A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of digital image processing, and in particular to a method and system for measuring the far-focus blur kernel of an optical imaging system. Background Technology
[0002] The blur kernel of an optical imaging system is generally used to describe the degree of blur in an image acquired by the system. Blur is largely caused by aberrations within the optical system. Aberrations can generally be divided into two categories: monochromatic aberrations, including spherical aberration, coma, astigmatism, field curvature, and distortion; and chromatic aberrations, caused by the different refractive indices of different wavelengths of monochromatic light in materials, including positional chromatic aberration and magnification chromatic aberration. In actual imaging, image degradation is usually caused by a variety of aberrations. To overcome lens aberrations and sharpen images, image restoration algorithms can be used to improve image quality by analyzing the characteristics of the optical imaging system.
[0003] In the field of digital image processing, image restoration algorithms are an important but challenging problem. For nearly half a century, image restoration has remained a hot research topic, presenting both theoretical challenges and significant practical applications. Image restoration, as a technique to improve image quality, estimates the original image based on observations of distortion and degradation, aiming to minimize or eliminate image distortion and noise. Image restoration methods can be categorized into non-blind convolutional image restoration and blind convolutional image restoration. Because image restoration is an ill-posed problem, blind convolutional image restoration algorithms are more ill-posed due to the greater number of unknown variables; therefore, non-blind restoration algorithms are more likely to achieve better restoration results. Accurate estimation of the blur kernel is crucial for achieving more ideal results with non-blind restoration algorithms. Currently, the blur kernel is typically obtained through calibration methods, i.e., using the optical system to be measured to acquire a calibration image, and then estimating the optical system blur kernel based on the difference between the actual captured image and the original calibration image using an algorithm.
[0004] The patented method for spatial variation PSF fusion estimation in single-lens computational imaging, which describes a calibration-based blur kernel measurement method, typically requires the calibration plate to fill the field of view of the optical system during calibration image acquisition. Furthermore, to acquire a calibration image with discernible details, focusing on the calibration plate is generally necessary. This results in the calibration plate being positioned relatively close to the optical system, allowing estimation of the blur kernel only at that focused location. In practical applications, the optical system often needs to focus on distant scenes for image acquisition. In such cases, the blur level of the optical system varies significantly, and the blur kernel obtained by the calibration-based blur kernel measurement method may not accurately reflect this blur level, thus affecting image restoration. Therefore, accurately estimating the blur kernel under long image distance conditions remains a worthy research problem. Summary of the Invention
[0005] In view of the above-mentioned technical deficiencies, the present invention provides a method and system for measuring the telephoto blur kernel of an optical imaging system, which improves the accuracy of lens blur kernel measurement and image restoration performance.
[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:
[0007] In a first aspect, a method for measuring the far-focus blur kernel of an optical imaging system includes the following steps:
[0008] S1. Simulate the illumination of stars at infinity using a collimator containing a point light source and frosted glass and a star plate, and collect blur kernel images of each area of the lens.
[0009] S2. A fuzzy kernel refining method based on central sub-connected regions refines fuzzy kernel images;
[0010] S3. Standardize the refined fuzzy kernel image.
[0011] Preferably, S1 includes:
[0012] The light from the point light source is passed through the frosted glass to generate a uniform surface light source; the surface light source is converted into concentrated light using a star plate; the light emitted from the star plate forms a blurred kernel image after entering the lens.
[0013] As a preferred embodiment, S1 also includes:
[0014] By applying the least squares circle fitting algorithm to the blurred kernel image, the center and radius of the blur spot in a region of the lens are located. A gridded sampling strategy is adopted, and the lens is moved according to the partition based on the lens field of view to collect blurred kernel images of each region within the field of view and record the blur spot radius of each region to construct a spatial variability mapping.
[0015] As a preferred option, point light sources include incandescent lamps or halogen lamps.
[0016] Preferably, S2 includes:
[0017] The blurred kernel image is binarized by setting non-zero pixels to 1 and the rest to 0 to generate a binary image. The center point of the binary image is set as the seed point, and non-zero pixels are marked as connected points. The connected points are traversed until no new connected points are added to generate a connected region. The corresponding region in the blurred kernel image is cropped based on the connected region to remove dark noise.
[0018] Preferably, S3 includes scaling the blur kernel image according to the actual size to maintain a uniform size for the blur kernel images in each region; and normalizing the blur kernel images in each region.
[0019] Preferably, S3 includes:
[0020] The pixel size of the blur kernel image is calculated using the number of photosensitive units, physical distance, and pixel size; let the number of photosensitive samples for each blur kernel be n1, n2, ..., n. t The physical sampling interval is l, the pixel size is m, and the pixel size of the blur kernel image is k. i Satisfy the following formula:
[0021] k i =l*m / n i , i = 1, 2, ..., t;
[0022] Based on pixel size and image size, the blur kernel image of each region is scaled to maintain uniform size.
[0023] Secondly, an optical imaging system for measuring far-focus blur kernels includes:
[0024] A collimator containing a point light source and frosted glass is used to generate a uniform surface light source.
[0025] A starlight plate is used in conjunction with a collimator to simulate starlight from infinity and to convert a surface light source into a focused beam of light. The light emitted from the starlight plate enters the lens and forms a blurred kernel image.
[0026] The fuzzy kernel refining module is used to refine fuzzy kernel images based on the fuzzy kernel refining method of central sub-connected regions.
[0027] The fuzz kernel standardization module is used to standardize the refined fuzz kernel image;
[0028] The aforementioned optical imaging system telefocal blur kernel measurement system is used to implement the optical imaging system telefocal blur kernel measurement method as described in the first aspect.
[0029] Thirdly, an electronic device is provided, including a processor and a memory;
[0030] The processor is connected to the memory;
[0031] The memory is used to store executable program code;
[0032] The processor runs a program corresponding to the executable program code stored in the memory to perform the method and steps described in the first aspect.
[0033] Fourthly, a computer-readable storage medium is provided having a computer program stored thereon, the computer-readable storage medium storing instructions that, when executed on a computer or processor, cause the computer or processor to perform the method and steps of the first aspect.
[0034] Compared with the prior art, the present invention has the following advantages and beneficial effects:
[0035] 1. Compared with the traditional calibration-based fuzzy kernel measurement method, the collimator-based fuzzy kernel measurement method can directly measure and acquire the actual fuzzy kernel image, thus more accurately understanding the degree of fuzziness. In contrast, the calibration-based method requires estimation of the fuzzy kernel through models and inference, which may introduce estimation errors.
[0036] 2. The fuzzy kernel measurement method based on collimator can adjust the measurement conditions according to actual needs, such as light source intensity, light source position and angle, while the calibration-based method requires obtaining the fuzzy kernel in advance through calibration images and models, and cannot directly adjust the degree of fuzziness.
[0037] 3. The collimator-based blur kernel measurement method can quickly acquire blur kernel images and provide real-time feedback on the blur effect, which is crucial for real-time image processing and adjusting blur parameters. In contrast, calibration-based methods require prior model calculation and parameter estimation, potentially necessitating more time and computational resources.
[0038] 4. The fuzzy kernel measurement method based on collimators does not require prior knowledge of the fuzzy kernel or calibration images; it can directly obtain fuzzy kernel information from the actual image. This has a significant advantage for some special scenarios or situations where calibration images cannot be obtained in advance.
[0039] In summary, the collimator-based fuzzy kernel measurement method has higher accuracy, controllability, and real-time performance compared to the calibration-based fuzzy kernel measurement method, and it does not require prior knowledge. This makes the collimator-based method a significant advantage and valuable application in fuzzy kernel measurement and real-time image processing. Attached Figure Description
[0040] Figure 1 Here is a flowchart of the method in Embodiment 1 of the present invention:
[0041] Figure 2 This is a schematic diagram of the parallel light tube measurement fuzzy kernel in Embodiment 1 of the present invention;
[0042] Figure 3 This is a flowchart of the fuzzy kernel refining method according to Embodiment 1 of the present invention;
[0043] Figure 4 This is a measurement fuzzy kernel example of Embodiment 1 of the present invention. Detailed Implementation
[0044] To make the technical means, inventive features, objectives, and effects of the invention readily understandable, the invention is further described below with reference to specific illustrations. However, the invention is not limited to the embodiments described below.
[0045] It should be noted that the structures, proportions, sizes, etc., illustrated in the accompanying drawings of this specification are only used to complement the content disclosed in the specification for those skilled in the art to understand and read, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.
[0046] This invention utilizes a collimator and a star-pointing plate to simulate starlight from infinity, collecting the responses of the starlight source at different imaging positions on an optical system to fit the point spread function of the imaging system at an infinity image distance. Then, based on the prior characteristics of the blur kernel distribution, a method based on central sub-connected region extraction is used to denoise the blur kernel, improving its accuracy. Finally, by estimating the pixel size of the actual blur kernel using the optical imaging system parameters, the blur kernel is standardized and normalized.
[0047] Example 1:
[0048] like Figure 1 The method for measuring the far-focus blur kernel in an optical imaging system, as shown, includes the following steps:
[0049] S1. Simulate the illumination of stars at infinity using a collimator containing a point light source and frosted glass and a star plate, and collect blur kernel images of each area of the lens.
[0050] Combining point light source measurement and a star chart, it is used to present a blur spot, i.e., a blur kernel image, at the lens. For example... Figure 2 As shown, the specific process is as follows: a collimator and a speckle plate are used to simulate starlight at infinity. The collimator contains a point light source and a frosted glass, and the light source is generally an incandescent or halogen lamp. A uniform surface light source is provided by the lamp and the frosted glass, and then the speckle plate is used to convert the surface light source into concentrated light. The light emitted from the collimator is affected by lens aberrations after entering the lens, ultimately forming a lens speckle image.
[0051] To ensure the geometric accuracy of the blur kernel and eliminate distortion caused by point source offset, a least-squares circle fitting algorithm is applied to the speckle image to accurately locate the center and radius of the speckle. This method can acquire the blur kernel of one region of the lens at a time. To obtain the blur kernel at various locations across the lens, a gridded sampling strategy is adopted for multi-location acquisition. The lens is moved according to partitions based on the lens's field of view to acquire the blur kernel of each region within the field of view, and the speckle radius at each location is recorded to construct a spatial variability mapping and reduce measurement errors caused by lens non-uniformity.
[0052] S2. Refine the blurred kernel image;
[0053] The fuzzy kernel refinement method based on central connected region extraction addresses the characteristics of low resolution and dark noise in fuzzy kernel images by using a central sub-connected region-based fuzzy kernel refinement method. This method sets the pixel values of non-zero regions outside the central sub-connected region to 0 in order to remove dark noise.
[0054] like Figure 3 As shown, the specific process of the fuzz kernel refinement method based on the extraction of the central connected region is as follows: First, the fuzz kernel image is binarized, and the pixels with non-zero pixel values are set to 1, while the rest are set to 0; then, the center point of the binary image is set as the seed point, and its non-zero pixels are marked as connected points. The newly added connected points are traversed, and the connected points in each direction are recorded; then the previous step is repeated until no new points are added; finally, based on the connected regions obtained in the previous step, the corresponding regions in the fuzz kernel are truncated to remove dark noise.
[0055] S3. Standardize the refined fuzzy kernel image;
[0056] Blur kernel standardization involves estimating the actual pixel size of the blur kernel through optical analysis. This size may differ from the directly acquired blur kernel image. The blur kernel image can be scaled based on its actual size to maintain uniformity across regions. Furthermore, the blur kernel must satisfy the condition that the sum of all its elements is 1; therefore, it undergoes normalization.
[0057] The specific process of blur kernel standardization is as follows: The pixel size of the blur kernel is estimated using the number of photosensitive units, physical distance, and pixel size. Let the number of photosensitive samples for each blur kernel be n1, n2, ..., n. t Given a sampling physical interval of l and a pixel size of m, the pixel size of the blur kernel is k. i Satisfy the following formula
[0058] k i =l*m / n i i = 1, 2, ..., t
[0059] Based on the measured pixel size of the blur kernel and the size of the acquired blur kernel image, the blur kernel images for each region are scaled to maintain uniform size. After this, to satisfy the prior condition that the integral sum equals 1, the blur kernel also needs to be normalized.
[0060] Compared with the prior art, the present invention has the following advantages and beneficial effects: The telephoto blur kernel measurement method of the optical imaging system proposed in the present invention improves the accuracy of lens blur kernel measurement and image restoration performance.
[0061] This patent proposes a method for measuring the blur kernel in a telephoto optical imaging system. Compared to traditional calibration-based blur kernel measurement methods, the collimator-based method can directly measure and acquire the actual blur kernel image, thus providing a more accurate understanding of the blur level. Calibration-based methods, on the other hand, require estimation of the blur kernel through models and inference, potentially introducing estimation errors. Secondly, the collimator-based method allows for adjustment of measurement conditions based on actual needs, such as light source intensity, position, and angle. Calibration-based methods, however, require prior acquisition of the blur kernel using calibration images and models, and cannot directly adjust the blur level. Thirdly, the collimator-based method can quickly acquire blur kernel images and provide real-time feedback on the blur effect, which is crucial for real-time image processing and blur parameter adjustment. In contrast, calibration-based methods require prior model calculations and parameter estimation, potentially requiring more time and computational resources. Finally, the collimator-based method does not require prior knowledge of the blur kernel or calibration images; it can directly obtain blur kernel information from the actual image. This offers significant advantages for special scenarios or situations where pre-acquiring calibration images is not possible. In summary, the collimator-based fuzzy kernel measurement method has higher accuracy, controllability, and real-time performance compared to the calibration-based fuzzy kernel measurement method, and it does not require prior knowledge. This makes the collimator-based method a significant advantage and valuable application in fuzzy kernel measurement and real-time image processing. Figure 4 The following is an example of an embodiment of the present invention.
[0062] Example 2:
[0063] An optical imaging system telephoto blur kernel measurement system, comprising:
[0064] A collimator containing a point light source and frosted glass is used to generate a uniform surface light source.
[0065] A starlight plate is used in conjunction with a collimator to simulate starlight from infinity and to convert a surface light source into a focused beam of light. The light emitted from the starlight plate enters the lens and forms a blurred kernel image.
[0066] The fuzzy kernel refining module is used to refine fuzzy kernel images based on the fuzzy kernel refining method of central sub-connected regions.
[0067] The fuzz kernel standardization module is used to standardize the refined fuzz kernel image;
[0068] Example 3:
[0069] From the description of the above embodiments, it is clear that each embodiment can be implemented by software and necessary general-purpose hardware platforms, or by hardware. Based on the understanding of these embodiments, the essential part of the technical solution can be embodied in a computer software product, stored in a read-write medium, such as a USB flash drive, external hard drive, ROM, RAM, disk, or optical disc. This software product includes several instructions for causing a computer device (such as a personal computer, server, or network device) to execute some of the methods of the above embodiments.
[0070] The foregoing is a detailed description of specific embodiments, but it does not limit the specific embodiments of the present invention to these descriptions. Without departing from the concept of the present invention, those skilled in the art can make appropriate substitutions or modifications to the described embodiments, which are considered effective within the scope of protection of the present invention.
Claims
1. A method for measuring the far-focus blur kernel in an optical imaging system, characterized in that, Includes the following steps: S1. Simulate the illumination of stars at infinity using a collimator containing a point light source and frosted glass and a star plate, and collect blur kernel images of each area of the lens. S2. A fuzzy kernel refining method based on central sub-connected regions refines fuzzy kernel images; S3. Standardize the refined fuzzy kernel image.
2. The method for measuring the far-focus blur kernel of an optical imaging system according to claim 1, characterized in that, S1 includes: light generated by a point light source is passed through frosted glass to generate a uniform surface light source; and the surface light source is converted into concentrated light using a star-shaped plate. The light emitted from the star plate forms a blurred kernel image after entering the lens.
3. The method for measuring the far-focus blur kernel of an optical imaging system according to claim 2, characterized in that, S1 also includes: using a least-squares circle fitting algorithm to locate the center and radius of the blur kernel image in a region of the lens; and using a gridded sampling strategy to move the lens according to the partitions based on the lens's field of view, acquiring blur kernel images of each region within the field of view and recording the blur kernel radius of each region to construct a spatial variability mapping.
4. The method for measuring the far-focus blur kernel of an optical imaging system according to claim 1, characterized in that, Point light sources include incandescent lamps or halogen lamps.
5. The method for measuring the far-focus blur kernel of an optical imaging system according to claim 1, characterized in that, S2 includes: Binarize the blurred kernel image by setting non-zero pixels to 1 and the rest to 0 to generate a binary image. Set the center point of the binary image as the seed point, mark the non-zero pixels as connected points, traverse the connected points until no new connected points are added, and generate a connected region; extract the corresponding region from the blurred kernel image based on the connected region to remove dark noise.
6. The method for measuring the far-focus blur kernel of an optical imaging system according to claim 1, characterized in that, S3 includes scaling the blur kernel image according to the actual size to maintain the uniform size of the blur kernel image in each region; and normalizing the blur kernel image in each region.
7. The method for measuring the far-focus blur kernel of an optical imaging system according to claim 6, characterized in that, S3 includes: calculating the pixel size of the blur kernel image using the number of photosensitive units, physical distance, and pixel size; and setting the number of photosensitive samples for each blur kernel as n1, n2, ..., n. t The physical sampling interval is l, the pixel size is m, and the pixel size of the blur kernel image is k. i Satisfy the following formula: k i =l*m / n i ,i=1,2,...,t; Based on pixel size and image size, the blur kernel image of each region is scaled to maintain uniform size.
8. A telephoto blur kernel measurement system for an optical imaging system, characterized in that, include: A collimator containing a point light source and frosted glass is used to generate a uniform surface light source. A starlight plate is used in conjunction with a collimator to simulate starlight from infinity and to convert a surface light source into a focused beam of light. The light emitted from the starlight plate enters the lens and forms a blurred kernel image. The fuzzy kernel refining module is used to refine fuzzy kernel images based on the fuzzy kernel refining method of central sub-connected regions. The fuzz kernel standardization module is used to standardize the refined fuzz kernel image; The aforementioned optical imaging system telefocal blur kernel measurement system is used to implement the optical imaging system telefocal blur kernel measurement method as described in claim 1.
9. An electronic device, characterized in that, Including the processor and memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code stored in the memory to perform the method and steps as described in claim 1.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program thereon, and the computer-readable storage medium stores instructions that, when the instructions are executed on a computer or processor, cause the computer or processor to perform the method and steps of claim 1.