A visual observation image quality evaluation device and method based on visual bionic structure

Abstract

The present application relates to the field of bionic vision, and particularly to a vision observation image quality evaluation device and method based on a vision bionic structure, which accurately simulates the independent diopter adjustment process of left and right eyes, and truly restores the imaging changes under myopia / hypermetropia and other diopter abnormalities; the consistency and fusion quality of binocular expected collection are improved through the establishment of an image fusion model, an evaluation system based on a deep integrated multi-modal fusion network model is established, the binocular vision bionic structure of the diopter state of the human eye is effectively simulated, left and right eyes are respectively focused, image collection and fusion processing are realized, so as to realize the real evaluation of the viewing effect of the vision bionic structure by the difference in the visual characteristics of the human eye.

Description

Technical Field

[0001] This invention relates to the field of bionic vision, and more particularly to a device and method for evaluating the quality of visual observation images based on a visual bionic structure. Background Technology

[0002] With the development of display technology, especially in fields such as virtual reality, augmented reality, intelligent detection, and bionic vision, the demand for simulating the imaging characteristics of human vision is increasing. Currently, there is no direct method for evaluating the effect of binocular visual observation; instead, analysis is performed using parameters such as brightness and contrast. However, differences in binocular refractive power among different populations are not reflected in the evaluation process, limiting its application in scenarios such as binocular vision bionics, refractive error simulation, and binocular image fusion.

[0003] In general, existing methods for evaluating the visual observation effects of binocular structures based on simulated human eye vision have the following shortcomings: First, they cannot simulate the impact of the difference in refractive power between the left and right eyes on the imaging focal length; second, there are no direct evaluation indicators, evaluation models, or evaluation methods for the impact of differences in binocular visual refractive power on imaging effects; third, there is a lack of a hardware structure and image processing algorithm coordination mechanism that simulates the visual accommodation ability of the human eye. Therefore, there is an urgent need for a binocular visual bionic structure that can simulate the refractive state of the human eye, enabling separate focusing, image acquisition, and fusion processing for the left and right eyes, in order to achieve a realistic evaluation of the viewing effect of the visual bionic structure on the differences in visual characteristics of the human eye. Summary of the Invention

[0004] To overcome the limitations of existing binocular vision mechanisms in simulating the effect of refractive error differences between the left and right eyes on imaging focal length, thus failing to provide a direct means of evaluating the impact of binocular vision refractive error differences on imaging effects, and unable to achieve separate focusing, image acquisition, and fusion processing for the left and right eyes, making it difficult to obtain a true assessment of the viewing effect of visual bionic structures, this invention provides a visual observation image quality assessment device based on visual bionic structures. The device includes:

[0005] Test image source, used to generate test images:

[0006] The bionic left eye optical path and the bionic right eye optical path have the same structure; the bionic left eye optical path and the bionic right eye optical path can independently set the test visual conditions for the left eye and the right eye, and acquire the corresponding left eye image and right eye image after imaging the test image at several observation distances, as well as acquire spectral correlation data corresponding to several field angles.

[0007] The database construction module is used to build a database that maps each observation distance to the test visual conditions of the left and right eyes, the left and right eye images, and spectral data.

[0008] The image fusion module is used to perform disparity-weighted fusion of the left-eye image and the right-eye image based on the database to obtain a fused image;

[0009] The evaluation module is used to evaluate the fused image and the spectral correlation data to obtain image quality evaluation results.

[0010] Preferably, both the bionic left eye optical path and the bionic right eye optical path include a variable aperture, a bionic visual lens, an imaging sensor, and a spectrometer; wherein,

[0011] The variable aperture is used to simulate the change in the size of the eye's pupil and change its own light-transmitting aperture.

[0012] The visual bionic lens is used to simulate the changes in the refractive state of the eye's lens to change its own focal length.

[0013] The imaging sensor is used to acquire an RGB image of the test image by simulating retinal imaging.

[0014] The spectrometer is used to collect the spectral brightness distribution at different viewing angles within the observation field area of ​​the bionic left eye optical path or the bionic right eye optical path.

[0015] Preferably, the bionic left eye optical path and the bionic right eye optical path are arranged sequentially along the light incident direction, with the variable aperture and the visual bionic lens being arranged in sequence.

[0016] The image light captured by the visual bionic lens is incident on the beam splitter via the first optical fiber to form two sub-image light rays; one sub-image light ray is incident on the imaging sensor via the second optical fiber, and the other sub-image light ray is incident on the spectrometer via the third optical fiber.

[0017] Preferably, the imaging sensors of the bionic left eye optical path and the bionic right eye optical path are connected to the image fusion module;

[0018] The spectrometers of the bionic left eye optical path and the bionic right eye optical path are connected to the database construction module.

[0019] Preferably, it also includes an output module for generating and outputting an evaluation report corresponding to the image quality assessment results based on the deep integrated multimodal fusion network model.

[0020] On the other hand, the present invention provides an image quality assessment method for a visual observation image quality assessment device based on a visual biomimetic structure, the method comprising the following steps:

[0021] S100: Independently set the test visual conditions for the left and right eyes; adjust the optical parameters of the bionic left eye optical path and the bionic right eye optical path;

[0022] S200: Acquire the left and right eye images corresponding to the bionic left eye optical path and the bionic right eye optical path respectively at several observation distances after imaging the test image, as well as the spectral correlation data corresponding to several field angles.

[0023] S300: Construct a database that maps each observation distance to the test visual conditions for the left and right eyes, the left and right eye images, and spectral data in a one-to-one correspondence.

[0024] S400: Perform disparity-weighted fusion of the left-eye and right-eye images according to the database to obtain a fused image;

[0025] S500: Evaluate the fused image and the spectral correlation data to obtain an image quality evaluation result;

[0026] S600: Generate and output an evaluation report corresponding to the image quality assessment results based on the deep integrated multimodal fusion network model.

[0027] Preferably, in S100, the test visual conditions for the left and right eyes are set independently, including:

[0028] Independently set the observation distance, refractive power, ambient brightness, and interpupillary distance for the left and right eyes respectively;

[0029] In S100, adjusting the optical parameters of the bionic left eye optical path and the bionic right eye optical path includes:

[0030] Adjust the aperture size of the variable aperture of the bionic left eye optical path and the bionic right eye optical path, as well as the focal length of the visual bionic lens.

[0031] Preferably, in step S200, spectral correlation data corresponding to several field angles at several observation distances are obtained for the bionic left eye optical path and the bionic right eye optical path, respectively. Specifically:

[0032] Obtain the maximum, minimum, and average values ​​of the spectral radiance of the bionic left and right eye optical paths at several viewing distances and at several field angles.

[0033] Preferably, in step S400, the left-eye image and the right-eye image are subjected to disparity-weighted fusion based on the database to obtain a fused image, specifically as follows:

[0034] The database is connected to a binocular vision image fusion model, and the left-eye and right-eye images are preprocessed using the binocular vision image fusion model to obtain the left-eye and right-eye images observed by real vision.

[0035] The binocular vision image fusion model is used to determine the overlapping area between the left-eye and right-eye images observed in real vision;

[0036] The disparity maps of the left-eye and right-eye images observed by real vision are obtained using the SAD algorithm; the disparity maps are then subjected to disparity-weighted fusion based on the overlapping regions to obtain a fused image.

[0037] Preferably, in step S500, the fused image and the spectral correlation data are evaluated to obtain an image quality assessment result, specifically as follows:

[0038] Construct an image fusion evaluation function and a spectral radiance evaluation function to generate a comprehensive evaluation function;

[0039] The fused image and the spectral correlation data are evaluated using the comprehensive evaluation function to obtain the image quality evaluation result.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] This invention achieves precise simulation of the independent refractive accommodation process of the left and right eyes, realistically reproducing the imaging changes under refractive abnormalities such as myopia / hyperopia; it improves the consistency and fusion quality of binocular pre-acquisition by establishing an image fusion model, and establishes an evaluation system based on a deep integrated multimodal fusion network model (DEMF) to effectively simulate the binocular visual bionic structure of the human eye's refractive state, realizing separate focusing, image acquisition, and fusion processing for the left and right eyes, so as to achieve a realistic evaluation of the viewing effect of the visual bionic structure on the differences in human eye visual characteristics, and provide a reliable technical means for testing stereo imaging systems and ophthalmic instruments. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0043] Figure 1 This is an overall structural diagram of a visual observation image quality assessment device based on a visual biomimetic structure provided by the present invention.

[0044] Figure 2 This is a structural block diagram of the bionic left eye optical path / bionic right eye optical path.

[0045] Figure 3 This is a light path diagram of the bionic left eye light path / bionic right eye light path.

[0046] Figure 4 This is a flowchart of a visual observation image quality assessment method based on visual biomimetic structures provided by the present invention.

[0047] Figure 5 This is the output DEMF display diagram.

[0048] Figure 6 It is a neural network layer of a deeply integrated multimodal fusion network model.

[0049] Figure reference numerals: 1. Test image source; 2. Variable aperture; 3. Visual bionic lens; 4. Beam splitter; 5. Imaging sensor; 6. Spectrometer; 7. First optical fiber; 8. Second optical fiber; 9. Third optical fiber. Detailed Implementation

[0050] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the present invention and not for limiting the present invention. Furthermore, it should be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all structures. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the present invention.

[0051] The terms "comprising" and "having," and any variations thereof, used in this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0052] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0053] Please see Figure 1 As shown, the present invention provides a visual observation image quality assessment device based on a visual biomimetic structure, the device comprising:

[0054] Test image source, used to generate test images: The test image source may be, but is not limited to, an LCD or LED display, capable of displaying images with different color layouts and / or displaying multi-color dynamic images;

[0055] The bionic left eye optical path and the bionic right eye optical path have the same structure; the aforementioned bionic left eye optical path and the aforementioned bionic right eye optical path are used to imitate the human left eye and right eye, and can independently image the aforementioned test image to obtain the corresponding left eye image and right eye image; the aforementioned bionic left eye optical path and the aforementioned bionic right eye optical path have the same optical path structure and the same monitoring component equipment layout, so that the aforementioned two optical paths have the function of simulating real left and right eye imaging.

[0056] The bionic left-eye optical path and the bionic right-eye optical path can independently set the test visual conditions for the left and right eyes. The test visual conditions may include, but are not limited to, observation distance, refractive power, ambient brightness, and interpupillary distance. By changing the test visual conditions of the bionic left-eye optical path and the bionic right-eye optical path, the optical parameters such as refractive power and focal length of the two bionic optical paths are made consistent with the optical parameters of the corresponding left and right eyes. Furthermore, it can realize the differentiated parameter settings of the bionic left-eye optical path and the bionic right-eye optical path, effectively forming the differentiation of binocular visual refractive power. At several observation distances, the corresponding left-eye image and right-eye image after imaging the test image are collected one by one, as well as the spectral correlation data corresponding to several field angles are collected one by one. At the same time, the bionic left-eye optical path and the bionic right-eye optical path are also equipped with imaging sensors and spectrometers respectively, so as to obtain the imaging images (RGB images) of the bionic left-eye optical path and the bionic right-eye optical path separately, and obtain the spectral brightness distribution of the left-eye image and the right-eye image separately.

[0057] The database construction module is used to build a database that maps each observation distance to the test visual conditions of the left and right eyes, the left and right eye images, and spectral data. The spectrometers of the bionic left and right eye optical paths are connected to the database construction module. The database construction module is used to map the test visual conditions of the left and right eyes, the left and right eye images, and the spectral data (spectral brightness distribution), providing a reliable basis for subsequent image fusion.

[0058] The image fusion module is used to perform disparity-weighted fusion of the left-eye and right-eye images according to the database to obtain a fused image; wherein, the imaging sensors of the bionic left-eye optical path and the bionic right-eye optical path are connected to the image fusion module.

[0059] The evaluation module is used to evaluate the fused image and spectral correlation data to obtain image quality evaluation results, thereby realizing the neural network evaluation of the fused image;

[0060] The output module is used to generate and output an evaluation report corresponding to the image quality assessment results based on the deep integrated multimodal fusion network model, thereby providing a comprehensive and accurate evaluation of binocular vision imaging.

[0061] Please see Figure 2-3As shown, both the bionic left eye optical path and the bionic right eye optical path include a variable aperture, a visual bionic lens, an imaging sensor, and a spectrometer. The system comprises: a variable aperture, which simulates changes in pupil size by altering its own aperture; a motorized aperture, which adjusts its aperture size as needed to precisely simulate pupil size changes; a bionic lens, which simulates changes in the refractive state of the eye's lens by altering its focal length; a zoom lens, which precisely simulates the refractive power of the left / right eye by changing its focal length to create a quasi-simulation of the binocular refractive power difference; an imaging sensor, which simulates retinal imaging by acquiring RGB images of the test image; a CMOS sensor or CCD sensor, which precisely forms an RGB color image; and a spectrometer, which acquires the spectral brightness distribution at different viewing angles within the field of view of the bionic left or right eye optical path. The spectrometer detects the brightness distribution of light in different wavelength ranges in the image light generated by the bionic lens, providing a reliable basis for subsequent image quality evaluation.

[0062] Furthermore, the bionic left eye optical path and the bionic right eye optical path are sequentially arranged with a variable aperture and a bionic visual lens along the incident light direction. The image light collected by the bionic visual lens is incident on a beam splitter via a first optical fiber to form two sub-image light rays. One sub-image light ray is incident on an imaging sensor via a second optical fiber, and the other sub-image light ray is incident on a spectrometer via a third optical fiber. The first optical fiber is used to transmit the image light rays generated by the bionic visual lens on the test image to the beam splitter. In this way, the beam splitter can preferentially split the received image light rays into two sub-image light rays of equal intensity. The second and third optical fibers are then used to transmit the corresponding sub-image light rays to the imaging sensor and the spectrometer, respectively, thereby realizing RGB imaging and spectral brightness detection.

[0063] Please see Figure 4 As shown, this invention provides a method for evaluating the quality of visual observation images based on visual biomimetic structures. The method includes the following steps:

[0064] S100: Independently set the test visual conditions for the left and right eyes; adjust the optical parameters of the bionic left eye optical path and the bionic right eye optical path;

[0065] S200: Acquire the left and right eye images corresponding to the bionic left eye optical path and the bionic right eye optical path respectively at several observation distances after imaging the test image, as well as the spectral correlation data corresponding to several field angles.

[0066] S300: Construct a database that maps each observation distance to the test visual conditions for the left and right eyes, the left and right eye images, and spectral data in a one-to-one correspondence.

[0067] S400: Perform disparity-weighted fusion of the left-eye and right-eye images based on the database to obtain a fused image;

[0068] S500: Evaluate the fused image and spectral correlation data to obtain image quality assessment results;

[0069] S600: Generates and outputs an evaluation report corresponding to the image quality assessment results based on the deep integrated multimodal fusion network model.

[0070] Furthermore, in S100, test visual conditions are independently set for the left and right eyes, including:

[0071] Independently set the observation distance, refractive power, ambient brightness, and interpupillary distance for the left and right eyes respectively;

[0072] In S100, the optical parameters of the bionic left eye optical path and the bionic right eye optical path are adjusted, including:

[0073] The aperture size of the variable aperture and the focal length of the bionic left and right eye optical paths are adjusted. Considering the differences in pupil size and refractive power between the left and right eyes of real humans, the optical parameters of the variable aperture and the bionic lens of the bionic left and right eye optical paths are adjusted to make the bionic left and right eye optical paths exhibit the same optical characteristics as the real left and right eyes.

[0074] Furthermore, in S200, spectral correlation data corresponding to several field angles at several observation distances are obtained for the bionic left eye optical path and the bionic right eye optical path, respectively. Specifically:

[0075] Obtain the maximum, minimum, and average values ​​of the spectral radiance of the bionic left and right eye optical paths at several viewing distances and at several field angles.

[0076] Specifically, after setting the aperture size of the variable apertures and the focal length of the bionic left and right eye optical paths, the distance between them (i.e., the interpupillary distance), and the ambient illuminance, the observation distance is changed to obtain the spectral brightness distribution (i.e., the maximum, minimum, and average values ​​of spectral radiance) corresponding to different field angles under a series of observation distance values ​​for each of the bionic left and right eye optical paths. The different field angles can be, but are not limited to, 1.7 mrad, 5 mrad, 11 mrad, and 100 mrad, and the maximum, minimum, and average values ​​of the spectral radiance corresponding to the above four field angles can be denoted as L. min、1.7mrad ( L max、1.7mrad ( L ave、1.7mrad ( ); L min、5mrad ( L max、5mrad ( L ave、5mrad ( ); L min、11mrad ( L max、11mrad ( L ave、11mrad ( ); L min、100mrad ( L max、100mrad ( L ave、100mrad ( ).

[0077] Furthermore, in S400, the left-eye and right-eye images are subjected to disparity-weighted fusion based on the database to obtain a fused image, specifically as follows:

[0078] The database is connected to the binocular vision image fusion model. The binocular vision image fusion model is used to preprocess the left-eye and right-eye images to obtain the left-eye and right-eye images observed by real vision.

[0079] The binocular vision image fusion model is used to determine the overlapping area between the left-eye and right-eye images observed in real vision;

[0080] The disparity maps of the left and right eye images observed by real vision are obtained using the SAD algorithm; the disparity maps are then weighted and fused based on the overlapping regions to obtain the fused image.

[0081] Specifically, the database is connected to a deep ensemble multimodal fusion network model (DEMF). DEMF is used to preprocess the left and right eye images by performing noise reduction filtering, feature extraction, distortion correction, two-dimensional Fourier transform, weighted processing based on the visual spatial sensitivity function (SFR), and inverse Fourier transform, to obtain the left and right eye images observed by real vision.

[0082] Next, the disparity maps of the left and right eye images observed by real vision are obtained using the SAD (Sum of Absolute Differences) algorithm; and the disparity maps are then weighted and fused based on the overlapping regions to obtain the fused image, as shown in the following formula:

[0083]

[0084] In the above formula, Represents the pixels of the merged image;

[0085] Represents the pixel values ​​of the left eye image; Represents the pixel values ​​of the right eye image; This represents the field-of-view weighting factor.

[0086] Furthermore, in S500, the fused image and spectral correlation data are evaluated to obtain image quality assessment results, specifically:

[0087] Construct an image fusion evaluation function and a spectral radiance evaluation function to generate a comprehensive evaluation function;

[0088] By using a comprehensive evaluation function, the fused image and spectral correlation data are evaluated to obtain the image quality assessment results.

[0089] Specifically, the process of constructing the image fusion evaluation function is as follows:

[0090] The first evaluation parameter is determined according to the following formula.

[0091]

[0092] in, Indicates the quantity of the first evaluation parameter; Indicates the modulation contrast of the test image; Indicates mean square error;

[0093]

[0094] in, This indicates the total number of pixels in the merged image; This indicates the number of pixels in the width direction of the merged image; Indicates the number of pixels in the height direction of the merged image; The x-axis represents the pixel index in the width direction of the merged image; the y-axis represents the pixel index in the height direction of the merged image. Represents the grayscale value of a pixel; This represents the standard grayscale value.

[0095] The second evaluation parameter is determined according to the following formula.

[0096]

[0097] in, Indicates the quantity of the second evaluation parameter; This represents the mean of binocular aberrations; This represents the standard deviation of the binocular aberrations after binocular image synthesis. This represents the mean grayscale value of the left eye image; This represents the mean grayscale value of the right eye image; This represents the standard deviation of the grayscale values ​​in the left eye image. This represents the standard deviation of the grayscale values ​​in the right eye image. and Represents the first and second constants.

[0098] Construct an image fusion evaluation function based on the following formula. ,

[0099] .

[0100] Specifically, the process of constructing the spectral radiance evaluation function is as follows:

[0101]

[0102] in, This represents the spectral radiance evaluation function under different field of view angles; The standard deviation of spectral radiance at different field of view angles; Indicates wavelength The intensity of light.

[0103] Specifically, the process of generating the comprehensive evaluation function is as follows:

[0104]

[0105] in, This represents the comprehensive evaluation function; Represents the image fusion evaluation function; This represents the spectral radiance evaluation function under different field of view angles; Indicates the ambient brightness corresponding to the left or right eye; Indicates the refractive power of the left or right eye; Indicates interpupillary distance; , , , , This represents the weighting coefficient.

[0106] Specifically, the image quality assessment results are as follows:

[0107]

[0108] in, This indicates the image quality assessment result; This represents the standard value under the current conditions. The corresponding output is... The image is shown below. Figure 5 As shown, the neural network layers of the deep integrated multimodal fusion network model are as follows: Figure 6 As shown.

[0109] The above image quality assessment results Image quality assessment results are directly proportional to image quality. The larger the value, the higher the image quality; the evaluation report may include, but is not limited to, basic parameter values ​​such as diopter, interpupillary distance, and illuminance settings, as well as image quality assessment results. And based on the above image quality assessment results The grading criteria, such as when If the image quality is good, then the image quality is excellent; when If the image quality is good, then the image quality is excellent; when If so, the image quality is poor.

[0110] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of a necessary general-purpose hardware platform, or by a combination of hardware and software. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a computer product. The present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Other embodiments may also be used. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A visual observation image quality evaluation apparatus based on visual biomimetic structure, characterized by, The device comprises: a test image source for generating a test image; a bionic left-eye optical path and a bionic right-eye optical path with the same structure; the bionic left-eye optical path and the bionic right-eye optical path can independently set left-eye and right-eye test visual conditions, and can independently collect left-eye images and right-eye images corresponding to the test image after imaging under a plurality of observation distances and a plurality of field angles respectively, and can independently collect a plurality of field angles corresponding to the spectral correlation data; the bionic left-eye optical path and the bionic right-eye optical path each comprise a variable diaphragm, a visual bionic lens, an imaging sensor, and a spectrometer; wherein the variable diaphragm is used to change its own light aperture by simulating the change of the pupil size of the eye; the visual bionic lens is used to change its own focal length by simulating the change of the refractive state of the lens of the eye; the imaging sensor is used to collect the RGB image after imaging of the test image by simulating the imaging of the retina of the eye; the spectrometer is used to collect the spectral luminance distribution under different field angles in the observation field of view region of the bionic left-eye optical path or the bionic right-eye optical path; a database construction module for constructing a database corresponding to each observation distance and the left-eye and right-eye test visual conditions, the left-eye images and the right-eye images, and the spectral correlation data; an image fusion module for performing disparity weighted fusion on the left-eye images and the right-eye images according to the database to obtain a fused image; an evaluation module for evaluating the fused image and the spectral correlation data to obtain an image quality evaluation result, specifically: constructing an image fusion evaluation function and a spectral radiation luminance evaluation function to generate a comprehensive evaluation function; evaluating the fused image and the spectral correlation data by using the comprehensive evaluation function to obtain the image quality evaluation result; an output module for generating and outputting an evaluation report corresponding to the image quality evaluation result according to the deep integration multi-modal fusion network model.

2. The device according to claim 1, wherein the bionic left-eye optical path and the bionic right-eye optical path are sequentially provided with the variable diaphragm and the visual bionic lens along the light incident direction; the image light collected by the visual bionic lens is incident to a light splitting prism through a first optical fiber to form two sub-image light rays; wherein one sub-image light ray is incident to the imaging sensor through a second optical fiber, and the other sub-image light ray is incident to the spectrometer through a third optical fiber.

3. The device according to claim 2, wherein the imaging sensor of each of the bionic left-eye optical path and the bionic right-eye optical path is connected to the image fusion module; the spectrometer of each of the bionic left-eye optical path and the bionic right-eye optical path is connected to the database construction module.

4. The image quality assessment method of the visual observation image quality assessment apparatus based on visual biomimetic structure according to any one of claims 1 to 3, characterized by, The method comprises the following steps: S100: independently setting the test visual conditions of the left eye and the right eye; adjusting the optical parameters of the bionic left-eye optical path and the bionic right-eye optical path; S200: acquiring the left-eye images and the right-eye images corresponding to the test image after imaging under a plurality of observation distances and a plurality of field angles corresponding spectral correlation data collected by the bionic left-eye optical path and the bionic right-eye optical path respectively; S300: constructing a database mapping each observation distance to left eye and right eye test visual conditions, left eye and right eye images, and spectral correlation data one-to-one; S400: disparity-weighted fusion of left eye and right eye images according to the database to obtain a fused image; S500: evaluation of the fused image and the spectral correlation data to obtain an image quality evaluation result; S600: generation and output of an evaluation report corresponding to the image quality evaluation result according to a deep integration multi-modal fusion network model.

5. The image quality evaluation method of claim 4, wherein in the S100, the test visual conditions of the left eye and the right eye are independently set, including: independently setting the observation distance, the diopter, the ambient brightness, and the interpupillary distance of the left eye and the right eye, respectively; in the S100, the optical parameters of the bionic left eye optical path and the bionic right eye optical path are adjusted, including: adjusting the aperture size of the variable diaphragm and the focal length of the visual bionic lens of the bionic left eye optical path and the bionic right eye optical path, respectively.

6. The image quality evaluation method of claim 4, wherein in the S200, the spectral correlation data of the bionic left eye optical path and the bionic right eye optical path one-to-one corresponding to a plurality of observation distances and a plurality of field angles are obtained, specifically: obtaining the maximum, minimum, and average values of the spectral radiance of the bionic left eye optical path and the bionic right eye optical path one-to-one corresponding to a plurality of observation distances and a plurality of field angles.

7. The image quality evaluation method of claim 4, wherein in the S400, the left eye and right eye images are disparity-weighted fused according to the database to obtain a fused image, specifically: connecting the database to a binocular vision image fusion model, pre-processing the left eye and right eye images using the binocular vision image fusion model to obtain left eye and right eye images observed by real vision; using the binocular vision image fusion model to determine the overlapping area of the left eye and right eye images observed by real vision; using the SAD algorithm to obtain a disparity map of the left eye and right eye images observed by real vision; and disparity-weighted fusing the disparity map according to the overlapping area to obtain a fused image.

8. The image quality evaluation method of claim 4, wherein in the S500, the fused image and the spectral correlation data are evaluated to obtain an image quality evaluation result, specifically: constructing an image fusion evaluation function and a spectral radiance evaluation function to generate a comprehensive evaluation function; using the comprehensive evaluation function to evaluate the fused image and the spectral correlation data to obtain an image quality evaluation result. ​ ​ ​ ​

Images