VR-based panoramic display system and panoramic image splicing method

Abstract

The invention relates to the technical field of panoramic images, in particular to a VR-based panoramic display system and a panoramic image splicing method, and the method comprises the following steps: obtaining external cause data of VR image equipment, the external cause data being a real-time pollution value of a lens; when the real-time pollution value of the lens meets the requirement, VR imaging is carried out; obtaining a VR image, and extracting feature points of textures according to the VR image; after the feature points of the texture are extracted, comparing and calculating the feature points of the real-time texture with the feature points of the known texture template, generating a real-time texture verification signal, and completing the recognition of the VR image; and obtaining a real-time texture verification signal, processing different verification results of the real-time texture based on the real-time texture verification signal, and finally completing VR image stitching. The invention provides a panoramic splicing scheme integrating pollution prevention and control, accurate feature matching and invalid data filtering, so that the automation degree and the output quality of a VR system are improved.

Description

Technical Field

[0001] The present invention relates to the field of panoramic image technology, and in particular to a VR-based panoramic display system and a panoramic image stitching method. Background Art

[0002] With the rapid development of virtual reality (VR) technology, panoramic display systems are increasingly being used in real estate, tourism, education and other fields. The quality of panoramic image stitching, a core component of VR, directly affects the user experience.

[0003] In existing technologies, mainstream stitching algorithms (such as SIFT and SURF) rely on general feature point extraction, but are not optimized for the complex texture structure of VR images: a single feature point model has difficulty distinguishing key topological structures such as texture endpoints and bifurcation points; the coupling calculation of position and orientation deviations is insufficient, resulting in a high mismatch rate (especially for rotated / scaled images), requiring manual secondary correction. Summary of the Invention

[0004] The purpose of the present invention is to provide a panoramic stitching solution that integrates pollution prevention and control, precise feature matching, and invalid data filtering, so as to improve the automation level and output quality of the VR system and solve the problems in the above background.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A VR-based panoramic image stitching method includes the following steps:

[0007] Step 1: Obtain the external data of the VR image device, which refers to the real-time contamination value of the lens;

[0008] When the real-time contamination value of the lens meets the requirements, VR imaging is performed;

[0009] Step 2: Obtain VR images and extract texture feature points based on the VR images;

[0010] Step 3: After the texture feature points are extracted, the feature points of the real-time texture are compared and calculated with the feature points of the known texture template to generate a real-time texture verification signal and complete the VR image recognition;

[0011] Step 4: Obtain a real-time texture verification signal. Based on the real-time texture verification signal, process different verification results of the real-time texture separately to finally complete VR image stitching.

[0012] As a further solution of the present invention: in step 2, the feature points of the texture include endpoints and bifurcation points;

[0013] Endpoints are the points in the texture where the lines begin and end;

[0014] A bifurcation point is the point where a line splits into two or more lines.

[0015] As a further solution of the present invention: the process of generating the real-time texture verification signal is:

[0016] The total value of the texture overall deviation is less than or equal to the total threshold of the texture overall deviation, and a real-time texture verification success signal is generated.

[0017] As a further solution of the present invention: the total value of the overall texture deviation is calculated by weighting the total value of the deviation between the endpoint of the texture template and the real-time texture endpoint and the total value of the deviation between the bifurcation point of the texture template and the real-time texture bifurcation point.

[0018] As a further solution of the present invention, the total deviation value between the texture template endpoint and the real-time texture endpoint is calculated by weighting the angle deviation value between the distance deviation value between the texture template endpoint and the real-time texture endpoint and the angle formed by the intersection of H and the x-axis and the angle formed by the intersection of h and the x-axis;

[0019] H is the path from the texture template endpoint to the coordinate origin, and h is the path from the real-time texture endpoint to the coordinate origin.

[0020] As a further solution of the present invention: the total deviation value between the texture template bifurcation point and the real-time texture bifurcation point is calculated by weighting the angle deviation value between the distance deviation value between the texture template bifurcation point and the real-time texture bifurcation point and the angle formed by the intersection of Q and the x-axis and the angle formed by the intersection of q and the x-axis;

[0021] Q is the path from the bifurcation point of the texture template to the coordinate origin, and q is the path from the bifurcation point of the real-time texture to the coordinate origin.

[0022] As a further solution of the present invention: the distance deviation value between the texture template endpoint and the real-time texture endpoint is calculated by the formula Ki=P-pi, where pi is the distance between the texture template endpoint and the real-time texture endpoint;

[0023] P is the maximum distance allowed between the texture template endpoint and the real-time texture endpoint;

[0024] According to the formula Li = |S-si|, the angular deviation between the angle formed by the intersection of H and the x-axis and the angle formed by the intersection of h and the x-axis is calculated;

[0025] S is the angle formed by the intersection of H and the x-axis, and si is the angle formed by the intersection of h and the x-axis.

[0026] As a further solution of the present invention: the angle deviation between the angle formed by the intersection of Q and the x-axis and the angle formed by the intersection of q and the x-axis is calculated by the formula Ni=|U-ui|,

[0027] U is the angle formed by the intersection of Q and the x-axis;

[0028] ui is the angle formed by the intersection of q and the x-axis.

[0029] As a further solution of the present invention: in step 4, based on the texture verification success signal, the cloud management and control platform obtains and processes the texture verification success signal, and marks the VR image as a preferred stitched panoramic image;

[0030] Based on the texture verification failure signal, the cloud management and control platform obtains and processes the texture verification success signal and marks the VR image as a non-preferred stitching panoramic image;

[0031] The obtained optimal stitched panoramic image is stitched together using an image stitching method.

[0032] A VR-based panoramic display system includes the following modules:

[0033] The initial inspection and identification module is used to obtain external factor data of VR imaging equipment. External factor data refers to the real-time contamination value of the lens;

[0034] When the real-time contamination value of the lens meets the requirements, VR imaging is performed;

[0035] Otherwise, VR imaging cannot be performed and the lens needs to be cleaned;

[0036] Feature point extraction module: obtains VR images and extracts texture feature points based on the VR images;

[0037] Feature point recognition and verification extraction module: After the feature points of the texture are extracted, the feature points of the real-time texture are compared and calculated with the feature points of the known texture template to generate a real-time texture verification signal to complete the recognition of the VR image;

[0038] Panoramic image verification module: obtains real-time texture verification signals, processes different verification results of real-time textures based on the real-time texture verification signals, and finally completes VR image stitching.

[0039] Beneficial effects of the present invention:

[0040] This invention uses an optical sensor to dynamically monitor the lens surface contamination value (Ti) and automatically compares it with a preset threshold (TI), effectively preventing image distortion caused by lens contamination. If contamination exceeds the limit, a cleaning prompt is immediately triggered, ensuring that subsequently acquired VR images are clear and usable, thereby improving panoramic image quality from the source. Texture feature points are refined into two key elements: endpoints and bifurcation points. A comprehensive deviation value is calculated through dual comparison of position and orientation. This mechanism significantly reduces the mismatch rate and ensures that the feature point comparison results are more consistent with the actual texture structure. VR images are intelligently graded based on texture verification results (success / failure signals): preferred stitched panoramic images (verification successful) are directly used for stitching, avoiding invalid data processing; non-preferred stitched panoramic images (verification failed) are automatically excluded, reducing system redundant operations. This strategy significantly improves the processing efficiency of valid images and conserves computing resources. Precise geometric transformations are performed on the preferred images (transformation matrices are calculated using feature points), and overlapping areas are processed using methods such as weighted averaging and maximum value selection, effectively eliminating brightness differences and seams. Ultimately, a unified coordinate transformation is used to generate a high-quality panoramic image with a coherent and natural visual effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present invention will be further described below with reference to the accompanying drawings.

[0042] Figure 1 This is a flow chart of a VR-based panoramic image stitching method provided by an embodiment of the present invention;

[0043] Figure 2 This is a flowchart of a VR-based panoramic display system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0045] Example 1

[0046] See also Figure 1 As shown, the present invention is a panoramic image stitching method based on VR, comprising the following steps:

[0047] Step 1: Obtain the external data of the VR image device, which refers to the real-time contamination value of the lens;

[0048] When the real-time contamination value of the lens meets the requirements, VR imaging is performed;

[0049] Otherwise, VR imaging cannot be performed and the lens needs to be cleaned;

[0050] Step 2: Obtain VR images and extract texture feature points based on the VR images;

[0051] Step 3: After the texture feature points are extracted, the feature points of the real-time texture are compared and calculated with the feature points of the known texture template to generate a real-time texture verification signal and complete the VR image recognition;

[0052] Step 4: Obtain a real-time texture verification signal. Based on the real-time texture verification signal, process different verification results of the real-time texture separately to finally complete VR image stitching.

[0053] In step 1, the process of obtaining the real-time lens contamination value is as follows:

[0054] The contamination on the lens surface is detected by an optical sensor. The optical sensor emits light and shines it onto the lens. The lens receives and reflects the light. The detector on the optical sensor receives the light reflected by the lens and converts it into an electrical signal. The optical sensor then uploads the electrical signal to the cloud management control platform to analyze the difference in reflected light signals between normal lenses and contaminated lenses. The standard reflected light signal of the lens is recorded as the lens surface contamination threshold, which is marked as TI. The real-time reflected light signal of the lens is recorded as the real-time lens contamination value, which is marked as Ti.

[0055] When Ti≤TI, it indicates that there is no contamination on the lens surface, which does not affect the normal image acquisition of the VR imaging device, and the VR imaging device can perform panoramic image acquisition;

[0056] When Ti>TI, it indicates that there is contamination on the lens surface, which affects the normal image acquisition of the VR imaging device. The VR imaging device cannot perform panoramic image acquisition, and the lens of the VR imaging device needs to be cleaned.

[0057] In step 2, the lens is scanned using an optical sensor, and the feature points of the texture are extracted by processing and analyzing the lens image;

[0058] The characteristic points of texture include endpoints and bifurcation points;

[0059] Endpoints refer to the starting and ending points of the texture, that is, the starting and ending points of a line;

[0060] A bifurcation point is the point where a line splits into two or more lines;

[0061] In step 3, the feature points of the texture include endpoints and bifurcation points. The positions of the real-time texture endpoints are compared with the positions of the texture template endpoints stored in the known texture image recognition device. The directions of the real-time texture endpoints are compared with the directions of the texture template endpoints stored in the known texture image recognition device to calculate the deviation values ​​of the endpoints.

[0062] Comparing the position of the real-time texture bifurcation point with the position of the texture template bifurcation point stored in the known texture image recognition device, and comparing the direction of the real-time texture bifurcation point with the direction of the texture template bifurcation point stored in the known texture image recognition device, and calculating the deviation value of the bifurcation point;

[0063] Establish a coordinate system with the center point of the texture image as the coordinate origin;

[0064] The endpoint coordinates of the texture template are marked as (XDi, YDi), (where i represents the value of the endpoint in the coordinate system, and i is a real number), (XDi represents the coordinate value of the texture template endpoint on the X axis, and YDi represents the coordinate value of the texture template endpoint on the Y axis);

[0065] The endpoint coordinates of the real-time texture are marked as (Xdi, Ydi), (where i represents the value of the endpoint in the coordinate system, and i is a real number), (Xdi represents the coordinate value of the real-time texture endpoint on the X axis, and Ydi represents the coordinate value of the real-time texture endpoint on the Y axis);

[0066] By formula Calculate the distance pi between the texture template endpoint and the real-time texture endpoint;

[0067] The maximum distance allowed between the texture template endpoint and the real-time texture endpoint is marked as P;

[0068] According to the formula Ki=P-pi, the distance deviation value Ki between the texture template endpoint and the real-time texture endpoint is calculated;

[0069] Take the x-axis of the coordinate system as the reference line;

[0070] The path from the texture template endpoint to the coordinate origin is recorded as Use the symbol H to represent (where i represents the value of the endpoint in the coordinate system, and i is a real number), (XDi represents the coordinate value of the texture template endpoint on the X axis, and YDi represents the coordinate value of the texture template endpoint on the Y axis);

[0071] The path from the real-time texture endpoint to the coordinate origin is recorded as Use the symbol h to represent (where i represents the value of the endpoint in the coordinate system, and i is a real number), (Xdi represents the coordinate value of the real-time texture endpoint on the X axis, and Ydi represents the coordinate value of the real-time texture endpoint on the Y axis);

[0072] The angle formed by the intersection of H and the x-axis is marked as S;

[0073] The angle formed by the intersection of h and the x-axis is marked as si;

[0074] According to the formula Li = |S-si|, the angle deviation value Li between the angle formed by the intersection of H and the x-axis and the angle formed by the intersection of h and the x-axis is calculated;

[0075] The total deviation value Zkl between the texture template endpoint and the real-time texture endpoint is calculated by the formula Zkl=c1*Ki+c2*Li, where c1 and c2 are weight coefficients, c1 is 0.52, and c2 is 0.48;

[0076] If the total deviation value Zkl between the texture template endpoint and the real-time texture endpoint is greater than the total deviation threshold between the texture template endpoint and the real-time texture endpoint, an endpoint verification failure signal is generated;

[0077] If the total deviation value Zkl between the texture template endpoint and the real-time texture endpoint is less than or equal to the total deviation threshold between the texture template endpoint and the real-time texture endpoint, an endpoint verification success signal is generated;

[0078] The bifurcation point coordinates of the texture template are marked as (XFe, YFe), (where e represents the value of the bifurcation point in the coordinate system, and e is a real number), (XFe represents the coordinate value of the bifurcation point of the texture template on the X axis, and YFe represents the coordinate value of the bifurcation point of the texture template on the Y axis);

[0079] The bifurcation point coordinates of the real-time texture are marked as (Xfe, Yfe), (where e represents the value of the bifurcation point in the coordinate system, and e is a real number), (XFe represents the coordinate value of the bifurcation point of the texture template on the X axis, and YFe represents the coordinate value of the bifurcation point of the texture template on the Y axis);

[0080] By formula Calculate the Euclidean distance bi between the texture template bifurcation point and the real-time texture bifurcation point;

[0081] The maximum distance allowed between the texture template bifurcation point and the real-time texture bifurcation point is marked as B;

[0082] According to the formula Mi=B-bi, the distance deviation value Mi between the texture template bifurcation point and the real-time texture bifurcation point is calculated;

[0083] The path from the bifurcation point of the texture template to the coordinate origin is recorded as Use the symbol Q to represent (where e represents the value of the bifurcation point in the coordinate system, and e is a real number), (XFe represents the coordinate value of the bifurcation point of the texture template on the X axis, and YFe represents the coordinate value of the bifurcation point of the texture template on the Y axis);

[0084] The path from the real-time texture bifurcation point to the coordinate origin is recorded as The symbol q represents (where e represents the value of the bifurcation point in the coordinate system, and e is a real number), (Xfe represents the coordinate value of the texture template bifurcation point on the X axis, and Yfe represents the coordinate value of the texture template bifurcation point on the Y axis);

[0085] The angle formed by the intersection of Q and the x-axis is marked as U;

[0086] The angle formed by the intersection of q and the x-axis is marked as ui;

[0087] According to the formula Ni = |U-ui|, the angle deviation Ni between the angle formed by the intersection of Q and the x-axis and the angle formed by the intersection of q and the x-axis is calculated;

[0088] The total deviation value Zmn between the texture template bifurcation point and the real-time texture bifurcation point is calculated by the formula Zmn=g1*Mi+g2*Ni, where g1 and g2 are weight coefficients, g1 is 0.52, and g2 is 0.48;

[0089] If the total deviation value Zmn between the texture template bifurcation point and the real-time texture bifurcation point is greater than the total deviation threshold between the texture template bifurcation point and the real-time texture bifurcation point, a bifurcation point verification failure signal is generated;

[0090] If the total deviation value Zmn between the texture template bifurcation point and the real-time texture bifurcation point is less than or equal to the total deviation threshold between the texture template bifurcation point and the real-time texture bifurcation point, a bifurcation point verification success signal is generated;

[0091] Obtaining a total deviation value Zkl between the endpoint of the texture template and the endpoint of the real-time texture, and a total deviation value Zmn between the bifurcation point of the texture template and the bifurcation point of the real-time texture;

[0092] The total value of texture deviation ZO is calculated by the formula ZO=a1*Zkl+a2*Zmn, where a1 and a2 are weight coefficients, a1 is 0.57, and a2 is 0.43;

[0093] If the texture overall deviation total value ZO is less than or equal to the texture overall deviation total threshold, it means that the real-time texture verification is successful, and a real-time texture verification success signal is generated;

[0094] If the texture overall deviation total value ZO is greater than or equal to the texture overall deviation total threshold, it means that the real-time texture verification fails, and a real-time texture verification failure signal is generated;

[0095] The real-time texture verification signal includes a real-time texture verification success signal and a real-time texture verification failure signal;

[0096] In step 4, the real-time texture verification signal includes a real-time texture verification success signal and a real-time texture verification failure signal;

[0097] Based on the texture verification success signal, the cloud management and control platform obtains and processes the texture verification success signal and marks the VR image as a preferred stitching panoramic image;

[0098] Based on the texture verification failure signal, the cloud management and control platform obtains and processes the texture verification success signal and marks the VR image as a non-preferred stitching panoramic image;

[0099] The obtained optimal stitched panoramic image is stitched together using an image stitching method, specifically:

[0100] The geometric transformation relationship between the images is calculated based on the matched feature points. The resulting transformation matrix is ​​then used to geometrically transform the input images so that they are aligned on the same plane. The aligned images are then merged into a single, larger image. This typically involves processing overlapping areas, such as addressing brightness inconsistencies and seams through methods like weighted averaging and maximum value selection. Based on the established mathematical transformation model, the images to be stitched are converted to the coordinate system of the reference image, completing a unified coordinate transformation. This allows multiple optimally stitched panoramic images to be seamlessly stitched into a single panoramic image.

[0101] The technical solution of the embodiment of the present invention: The present invention completes the recognition of the texture image of the smart lock by comparing and calculating the feature points of the real-time texture with the feature points of the known texture template, that is, by comparing and calculating the position of the real-time texture endpoint and the known texture template endpoint, comparing and calculating the direction of the real-time texture endpoint and the known texture template endpoint, comparing and calculating the position of the real-time texture bifurcation point and the known texture template bifurcation point, and comparing and calculating the direction of the real-time texture bifurcation point and the known texture template bifurcation point, obtaining a real-time texture verification signal, and the obtained real-time texture verification signal reflects the verification result of the real-time texture. Different verification results of the real-time texture are processed separately, and finally VR image stitching is completed. The present invention dynamically monitors the lens surface contamination value (Ti) through an optical sensor and automatically compares it with a preset threshold (TI), effectively avoiding image distortion caused by lens contamination. If the contamination exceeds the standard, a cleaning prompt is immediately triggered to ensure that the subsequently collected VR image is clear and usable, thereby improving the panoramic image quality from the source. The texture feature points are refined into two key elements: endpoints and bifurcation points, and the comprehensive deviation value is calculated through dual comparison of position and direction. This mechanism significantly reduces the mismatch rate and ensures that the feature point comparison results are more consistent with the actual texture structure. Based on the texture verification results (success / failure signals), VR images are intelligently graded: preferred stitched panoramic images (verification successful): directly used for stitching to avoid invalid data processing; non-preferred stitched panoramic images (verification failed): automatically excluded to reduce system redundant operations. This strategy greatly improves the processing efficiency of valid images and saves computing resources. Precise geometric transformation is performed on the preferred images (the transformation matrix is ​​calculated through feature points), and weighted average / maximum selection and other methods are used to process overlapping areas, effectively eliminating brightness differences and seam problems. Finally, a high-quality panoramic image is generated through unified coordinate transformation, and the visual effect is coherent and natural.

[0102] Example 2

[0103] See also Figure 2 As shown, based on the above embodiment 1, the present invention is a VR-based panoramic display system, including the following modules:

[0104] The initial inspection and identification module is used to obtain external factor data of VR imaging equipment. External factor data refers to the real-time contamination value of the lens;

[0105] When the real-time contamination value of the lens meets the requirements, VR imaging is performed;

[0106] Otherwise, VR imaging cannot be performed and the lens needs to be cleaned;

[0107] Feature point extraction module: obtains VR images and extracts texture feature points based on the VR images;

[0108] Feature point recognition and verification extraction module: After the feature points of the texture are extracted, the feature points of the real-time texture are compared and calculated with the feature points of the known texture template to generate a real-time texture verification signal to complete the recognition of the VR image;

[0109] Panoramic image verification module: obtains real-time texture verification signals, processes different verification results of real-time textures based on the real-time texture verification signals, and finally completes VR image stitching.

[0110] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A panoramic image stitching method based on VR, characterized in that: The following steps are involved: Step 1: Obtain the external data of the VR image device, which refers to the real-time contamination value of the lens; When the real-time contamination value of the lens meets the requirements, VR imaging is performed; Step 2: Obtain VR images and extract texture feature points based on the VR images; Step 3: After the texture feature points are extracted, the feature points of the real-time texture are compared and calculated with the feature points of the known texture template to generate a real-time texture verification signal and complete the VR image recognition; Step 4: Obtain a real-time texture verification signal. Based on the real-time texture verification signal, process different verification results of the real-time texture separately to finally complete VR image stitching.

2. The VR-based panoramic image stitching method according to claim 1, characterized in that: In step 2, the feature points of the texture include endpoints and bifurcation points; Endpoints are the points in the texture where the lines begin and end; A bifurcation point is the point where a line splits into two or more lines.

3. The VR-based panoramic image stitching method according to claim 2, characterized in that: The generation process of real-time texture verification signal is as follows: The total value of the texture overall deviation is less than or equal to the total threshold of the texture overall deviation, and a real-time texture verification success signal is generated.

4. The VR-based panoramic image stitching method according to claim 3, characterized in that: The total value of the overall texture deviation is calculated by the total value of the deviation between the endpoint of the texture template and the endpoint of the real-time texture and the total value of the deviation between the bifurcation point of the texture template and the bifurcation point of the real-time texture.

5. The VR-based panoramic image stitching method according to claim 4, characterized in that: The total deviation between the texture template endpoint and the real-time texture endpoint is calculated as follows: the total deviation between the texture template endpoint and the real-time texture endpoint is weighted by the angle deviation between the angle formed by the intersection of H and the x-axis and the angle formed by the intersection of h and the x-axis; H is the path from the texture template endpoint to the coordinate origin, and h is the path from the real-time texture endpoint to the coordinate origin.

6. The VR-based panoramic image stitching method according to claim 5, characterized in that: The total deviation value between the texture template bifurcation point and the real-time texture bifurcation point is calculated by weighting the angle deviation value between the distance deviation value between the texture template bifurcation point and the real-time texture bifurcation point and the angle formed by the intersection of Q and the x-axis and the angle formed by the intersection of q and the x-axis; Q is the path from the bifurcation point of the texture template to the coordinate origin, and q is the path from the bifurcation point of the real-time texture to the coordinate origin.

7. The VR-based panoramic image stitching method according to claim 6, characterized in that: The distance deviation between the texture template endpoint and the real-time texture endpoint is calculated by the formula Ki=P-pi, where pi is the distance between the texture template endpoint and the real-time texture endpoint; P is the maximum distance allowed between the texture template endpoint and the real-time texture endpoint; According to the formula Li = |S-si|, the angular deviation between the angle formed by the intersection of H and the x-axis and the angle formed by the intersection of h and the x-axis is calculated; S is the angle formed by the intersection of H and the x-axis, and si is the angle formed by the intersection of h and the x-axis.

8. The VR-based panoramic image stitching method according to claim 7, characterized in that: The angle deviation between the angle formed by the intersection of Q and the x-axis and the angle formed by the intersection of q and the x-axis is calculated by the formula Ni=|U-ui|. U is the angle formed by the intersection of Q and the x-axis; ui is the angle formed by the intersection of q and the x-axis.

9. The VR-based panoramic image stitching method according to claim 1, characterized in that: In step 4, based on the texture verification success signal, the cloud management and control platform obtains and processes the texture verification success signal and marks the VR image as a preferred stitching panoramic image; Based on the texture verification failure signal, the cloud management and control platform obtains and processes the texture verification success signal and marks the VR image as a non-preferred stitching panoramic image; The obtained optimal stitched panoramic image is stitched together using an image stitching method.

10. A VR-based panoramic display system, characterized in that: Includes the following modules: The initial inspection and identification module is used to obtain external factor data of VR imaging equipment. External factor data refers to the real-time contamination value of the lens; When the real-time contamination value of the lens meets the requirements, VR imaging is performed; Otherwise, VR imaging cannot be performed and the lens needs to be cleaned; Feature point extraction module: obtains VR images and extracts texture feature points based on the VR images; Feature point recognition and verification extraction module: After the feature points of the texture are extracted, the feature points of the real-time texture are compared and calculated with the feature points of the known texture template to generate a real-time texture verification signal to complete the recognition of the VR image; Panoramic image verification module: obtains real-time texture verification signals, processes different verification results of real-time textures based on the real-time texture verification signals, and finally completes VR image stitching.

Images