Image restoration method, system and storage module

Abstract

An image restoration method, system, and storage module are disclosed. The method comprises: obtaining an original image and shooting information of the original image, the shooting information including a shooting position and a shooting angle; the shooting position being the position of a camera device when shooting the original image, and the shooting angle including the wide angle and pitch angle of the camera device when shooting the original image; obtaining a landmark feature that can be calibrated with geographic coordinates from the original image, obtaining pixels corresponding to the landmark feature as feature pixels, and marking the geographic coordinates corresponding to the landmark feature as reference coordinates; and calculating the depth of field of the feature pixel using the shooting position and the reference coordinates as a reference depth of field. The present invention ensures that the depth of field of the feature pixel is obtained quickly, conveniently, efficiently, and accurately, laying the foundation for the subsequent accurate calculation of the depth of field of the remaining pixels in the original image, thereby ensuring high-quality restoration of the original image.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to an image restoration method, system and storage module. Background Art

[0002] As image processing technology is widely used in various industries, people's pursuit of image quality is getting higher and higher. In this context, research on image restoration technology is endless.

[0003] The most recommended image restoration technology currently combines pixel depth of field with a pre-built image restoration model to restore the original image. Currently, pixel depth of field can be obtained using two methods: manually measuring the distance from the shooting position to certain pixels in the original image, which is the depth of field; and estimating the distance from the shooting position to certain pixels in the original image, which is the depth of field. The former method is difficult, time-consuming, and difficult to implement; the latter method suffers from large errors and cannot guarantee image restoration quality. Summary of the Invention

[0004] In order to solve the defect in the prior art that it is difficult to obtain the depth of field of pixel points on the captured original image, the present invention proposes an image restoration method, system and storage module.

[0005] One of the objectives of the present invention is to provide an image restoration method that can accurately and quickly obtain the depth of field of pixels on the original image, thereby ensuring the implementation and quality of image restoration.

[0006] An image restoration method comprises the following steps:

[0007] S1. Obtaining an original image and shooting information of the original image, wherein the shooting information includes a shooting position and a shooting angle; the shooting position is the position of the camera when the original image is shot, and the shooting angle includes a wide angle and a pitch angle when the camera shoots the original image;

[0008] S2. Obtaining landmark features that can be calibrated with geographic coordinates from the original image, obtaining pixel points corresponding to the landmark features as feature pixels, and marking the geographic coordinates corresponding to the landmark features as reference coordinates; and calculating the depth of field of the feature pixels by combining the shooting positioning and the reference coordinates as the reference depth of field;

[0009] S3, calculating the depth of field of each pixel on the original image by combining the shooting angle and the reference depth of field;

[0010] S4. Restoring the original image by combining the image restoration model with the depth of field of each pixel on the original image.

[0011] Preferably, in S2, the reference coordinates are obtained by obtaining image features existing on the GIS map from the original image as landmark features, and obtaining geographic coordinates corresponding to the landmark features in combination with the GIS map as reference coordinates; landmark features include buildings and natural attractions.

[0012] Preferably, the shooting information also includes weather information, and S4 includes the following sub-steps:

[0013] S41, setting meteorological categories and constructing image restoration models corresponding to the meteorological categories;

[0014] S42, obtaining the meteorological category to which the meteorological information in the shooting information belongs, and obtaining an image restoration model corresponding to the meteorological category as a target model;

[0015] S43. Restoring the original image by combining the target model with the depth of field of each pixel on the original image.

[0016] Preferably, the meteorological category includes weather conditions and time conditions. The weather conditions include foggy days, rainy days, sunny days and snowy days; and the time conditions include daytime and nighttime.

[0017] Preferably, in S1, the method for obtaining meteorological information of the original image includes the following steps:

[0018] S11, obtaining the shooting time and shooting location associated with the original image;

[0019] S12, combining the shooting time and the shooting location network to obtain local time conditions and weather conditions;

[0020] S13. Generate meteorological information based on local time conditions and weather conditions.

[0021] A second object of the present invention is to provide an image restoration system suitable for the above-mentioned image restoration method.

[0022] An image restoration system includes: a camera device and a processor;

[0023] A camera device, configured to capture an original image, wherein the original image is associated with shooting information; the shooting information includes a shooting position and a shooting angle; the shooting position is the position of the camera device when capturing the original image, and the shooting angle is the wide angle and pitch angle of the camera device when capturing the original image;

[0024] The processor is used to obtain the original image captured by the camera device, and the processor is also used to process the original image according to the above-mentioned image restoration method.

[0025] Preferably, it further comprises a model storage module, wherein the model storage module is used to store image restoration models corresponding one to one to meteorological categories.

[0026] Preferably, it further comprises a parameter editing module, which is used to manually set landmark features and reference coordinates.

[0027] The third object of the present invention is to provide a storage module that is conducive to the promotion of the above-mentioned image restoration method.

[0028] A storage module stores a computer program, and when the computer program is executed, it is used to implement the above-mentioned image restoration method.

[0029] The advantages of the present invention are:

[0030] (1) The image restoration method proposed in the present invention combines the shooting positioning and the geographic coordinates of the landmark features to obtain the depth of field of the characteristic pixel points in the original image, ensuring that the depth of field of the characteristic pixel points is obtained quickly, conveniently, efficiently and accurately, laying the foundation for the subsequent accurate calculation of the depth of field of the remaining pixel points on the original image, thereby providing a guarantee for high-quality restoration of the original image.

[0031] (2) The original image is automatically identified with the GIS map, which ensures the objectivity of the extraction of the landmark features and the accurate correspondence between the landmark features and the reference coordinates.

[0032] (3) Corresponding image restoration models are established for different meteorological categories, which enables targeted restoration of original images based on shooting conditions, further improving the quality of image restoration.

[0033] (4) Meteorological categories include weather conditions and time conditions. The time conditions corresponding to the original image can be determined based on the shooting time and shooting location. Taking into account the time difference, the image restoration model is further accurately selected according to the shooting conditions of the original image, thereby ensuring the image restoration quality.

[0034] (5) The present invention proposes an image restoration system that can achieve high-quality restoration of original images captured by a camera. The system stores image restoration models that correspond to meteorological categories. This allows the processor to directly invoke the stored image restoration models when restoring the original image, improving image processing efficiency.

[0035] (6) The image restoration system also includes a parameter editing module. In this way, when the landmark features or the geographical coordinates of the landmark features cannot be determined, the staff can manually set the landmark features and reference coordinates through the parameter editing module to ensure that the system can work in an emergency and improve work reliability.

[0036] (7) The storage medium proposed in the present invention realizes plug-and-play by writing the image restoration method into a storage module in the form of a computer program, which is conducive to the promotion of the image restoration method. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a flow chart of an image restoration method;

[0038] Figure 2 This is a flow chart of another image restoration method;

[0039] Figure 3 Flowchart of the method for obtaining meteorological information from original images. DETAILED DESCRIPTION

[0040] Reference Figure 1 , an image restoration method proposed in this embodiment includes the following steps:

[0041] S1. Obtain an original image and shooting information of the original image, wherein the shooting information includes a shooting position and a shooting angle. The shooting position refers to the position of the camera when the original image was captured, and the shooting angle includes the wide angle and the pitch angle when the camera captured the original image. The shooting position and shooting angle are automatically recorded by the camera.

[0042] S2. Obtain landmark features that can be calibrated with geographic coordinates from the original image, obtain pixel points corresponding to the landmark features as feature pixels, and mark the geographic coordinates corresponding to the landmark features as reference coordinates; combine the shooting positioning and the reference coordinates to calculate the depth of field of the feature pixels as the reference depth of field.

[0043] S3, calculating the depth of field of each pixel on the original image by combining the shooting angle and the reference depth of field;

[0044] S4. Restoring the original image by combining the image restoration model with the depth of field of each pixel on the original image.

[0045] In step S2, the number of feature pixels is greater than one. Specifically, the feature pixels can be set based on the actual size of the space covered by the original image. The larger the actual space, the greater the number of feature pixels. In this embodiment, the number of feature pixels is at least three to more clearly extract the spatial distance in the original image, thereby ensuring the accuracy of the depth of field of the remaining pixels in the original image calculated by combining the shooting angle and the reference depth of field. To further ensure the accuracy of the depth of field calculation, the multiple feature pixels selected should have different depths of field.

[0046] It is worth noting that in step S3, when the reference depth of field and shooting angle are known, the depth of field calculation of the remaining pixels on the original image (i.e., pixels other than the feature pixels) can be implemented using existing technology, so it will not be explained in detail here.

[0047] In step S4, the input of the image restoration model is the original image and the depth of field of each pixel on the original image, and the output of the image restoration model is the restored image. In specific implementation, the image restoration model can adopt an existing known image restoration model.

[0048] During specific implementation, an image restoration model can also be obtained based on neural network training. The specific steps are: obtaining training samples, which are composed of the original image taken and the depth of field of each pixel on the original image; obtaining the restored image corresponding to each training sample as the label of the training sample, and then learning the training samples and the corresponding labels based on the neural network model to obtain an image restoration model with the original image and the depth of field of each pixel on the original image as input and the restored image as output.

[0049] In this embodiment, the depth of field of the characteristic pixel points in the original image is obtained by combining the shooting positioning and the geographic coordinates of the landmark features, which ensures that the depth of field of the characteristic pixel points is obtained quickly, conveniently, efficiently and accurately, laying the foundation for the subsequent accurate calculation of the depth of field of the remaining pixel points on the original image, thereby providing a guarantee for high-quality restoration of the original image.

[0050] In step S2, reference coordinates are obtained by extracting image features present on the GIS map from the original image as landmark features, and then obtaining the geographic coordinates corresponding to these landmark features in conjunction with the GIS map as reference coordinates. Landmark features include buildings and natural attractions. Thus, in this embodiment, landmark features in the original image are automatically identified in conjunction with the GIS map, ensuring the objectivity of landmark feature extraction and, consequently, the precise correspondence between these landmark features and the reference coordinates.

[0051] In specific implementation, after obtaining the original image, the shooting area can be determined in combination with the shooting location and shooting angle associated with the original image, and then the model within the shooting area can be enlarged on the GIS map as a real-scene model. Then, the image features that match the original image and the real-scene model are determined through existing image comparison technology as screening features, and the screening features that can determine the geographic coordinates are obtained in combination with the real-scene model as landmark features, thereby obtaining the reference coordinates of the landmark features.

[0052] In this embodiment, the image features in the original image are screened and located in combination with the GIS map, ensuring the accurate identification and positioning of the landmark features, thereby ensuring the intelligent, accurate and reliable acquisition of feature pixel points by combining reference coordinates and shooting positioning.

[0053] Reference Figure 2 In this embodiment, the shooting information also includes weather information, and S4 includes the following sub-steps.

[0054] S41. Set meteorological categories and construct image restoration models corresponding to the meteorological categories.

[0055] S42: Obtain the meteorological category to which the meteorological information in the shooting information belongs, and obtain an image restoration model corresponding to the meteorological category as a target model.

[0056] S43. Restoring the original image by combining the target model with the depth of field of each pixel on the original image.

[0057] In specific implementations, meteorological categories include weather conditions and time conditions. Weather conditions include foggy, rainy, sunny, and snowy; time conditions include daytime and nighttime. Thus, in this embodiment, by combining weather conditions and time conditions, eight meteorological categories can be obtained: foggy daytime, rainy daytime, sunny daytime, snowy daytime, foggy nighttime, rainy nighttime, sunny nighttime, and snowy nighttime. In specific implementations, weather conditions can be further categorized. For example, rainy days can be categorized into light rain, moderate rain, heavy rain, and rainstorms. Thus, more refined weather conditions can yield even more refined meteorological categories, such as light rain daytime, moderate rain daytime, heavy rain daytime, and rainstorm daytime.

[0058] In this embodiment, corresponding image restoration models are established for different meteorological categories, thereby achieving targeted restoration of the original image according to the shooting conditions, and further improving the quality of image restoration.

[0059] In specific implementation, meteorological information can be obtained based on weather forecast data. Figure 3 In step S1, the method for obtaining meteorological information of the original image includes the following steps.

[0060] S11 , obtaining the shooting time and shooting location associated with the original image.

[0061] S12: combining the shooting time and the shooting location network to obtain local time conditions and weather conditions.

[0062] S13. Generate meteorological information based on local time conditions and weather conditions.

[0063] It is worth noting that the weather conditions in step S12 can be classified in the same way as the weather conditions in the meteorological category corresponding to the image restoration model, for example, the weather conditions are both classified as foggy, rainy, sunny, and snowy; or the weather conditions in step S12 can be classified in a more specific way than the weather conditions in the meteorological category corresponding to the image restoration model, for example, the weather conditions in the meteorological category corresponding to the image restoration model are classified as foggy, rainy, sunny, and snowy, while the weather conditions in step S12 are classified as light fog, heavy fog, light rain, moderate rain, heavy rain, rainstorm, cloudy, sunny, light snow, moderate snow, moderate snow, and heavy snow with reference to the weather forecast. In this way, the corresponding meteorological category can be accurately locked based on the meteorological information of the original image.

[0064] In this embodiment, the time conditions corresponding to the original image can also be determined based on the shooting time and shooting location, taking into account the time difference, further ensuring that the image restoration model is accurately selected according to the shooting conditions of the original image, thereby ensuring the image restoration quality.

[0065] In this embodiment, an image restoration system includes: a camera device and a processor.

[0066] A camera device is used to capture an original image, and the original image is associated with shooting information; the shooting information includes a shooting position and a shooting angle; the shooting position is the position of the camera device when capturing the original image, and the shooting angle is the wide angle and pitch angle of the camera device when capturing the original image.

[0067] Specifically, the shooting positioning is provided by a positioning module provided by the camera device, and the shooting angle is provided by an angle sensor provided by the camera device.

[0068] The processor is configured to obtain an original image captured by a camera device and to process the original image according to the above-described image restoration method. Specifically, after obtaining the original image associated with the capture information, the processor obtains a landmark feature from the original image as a feature pixel, and uses a GIS map to obtain the geographic coordinates of the landmark feature as reference coordinates corresponding to the feature pixel. The processor then calculates the distance between the capture location and the reference coordinates as the depth of field of the feature pixel, which serves as a reference depth of field. The processor then combines the capture angle and the reference depth of field to obtain the depth of field of each remaining pixel in the original image. The processor can then restore the original image by combining the depth of field of each pixel in the original image with an image restoration model.

[0069] The image restoration system also includes a model storage module for storing image restoration models that correspond one-to-one with meteorological categories. This allows the processor to directly call the stored image restoration model when restoring the original image, improving image processing efficiency.

[0070] The image restoration system also includes a parameter editing module for manually setting landmark features and reference coordinates. This allows staff to manually set landmark features and reference coordinates when landmark features or their geographic coordinates are unavailable, ensuring the system's emergency response and improving reliability.

[0071] This embodiment also provides a storage module that stores a computer program. When executed, the computer program is used to implement the aforementioned image restoration method. Thus, in this embodiment, by writing the image restoration method into the storage module as a computer program, plug-and-play is achieved, facilitating the widespread use of the image restoration method.

[0072] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An image restoration method, characterized in that: The following steps are involved: S1. Obtaining an original image and shooting information of the original image, wherein the shooting information includes a shooting position and a shooting angle; the shooting position is the position of the camera when the original image is shot, and the shooting angle includes a wide angle and a pitch angle when the camera shoots the original image; S2. Obtaining landmark features that can be calibrated with geographic coordinates from the original image, obtaining pixel points corresponding to the landmark features as feature pixels, and marking the geographic coordinates corresponding to the landmark features as reference coordinates; and calculating the depth of field of the feature pixels by combining the shooting positioning and the reference coordinates as the reference depth of field; S3, calculating the depth of field of each pixel on the original image by combining the shooting angle and the reference depth of field; S4, restoring the original image by combining the depth of field of each pixel on the original image with the image restoration model; In S2, the reference coordinates are obtained by obtaining image features existing on the GIS map from the original image as landmark features, and obtaining the geographic coordinates corresponding to the landmark features in combination with the GIS map as reference coordinates; landmark features include buildings and natural attractions; The shooting information also includes weather information. S4 includes the following sub-steps: S41, setting meteorological categories and constructing image restoration models corresponding to the meteorological categories; S42, obtaining the meteorological category to which the meteorological information in the shooting information belongs, and obtaining an image restoration model corresponding to the meteorological category as a target model; S43. Restoring the original image by combining the target model with the depth of field of each pixel on the original image.

2. The image restoration method according to claim 1, wherein: Meteorological categories include weather conditions and time conditions. Weather conditions include: foggy, rainy, sunny and snowy; time conditions include: day and night.

3. The image restoration method according to claim 2, wherein: In S1, the method for obtaining meteorological information of the original image includes the following steps: S11, obtaining the shooting time and shooting location associated with the original image; S12, combining the shooting time and the shooting location network to obtain local time conditions and weather conditions; S13. Generate meteorological information based on local time conditions and weather conditions.

4. An image restoration system, characterized in that: include: Camera device and processor; A camera device, configured to capture an original image, wherein the original image is associated with shooting information; the shooting information includes a shooting position and a shooting angle; the shooting position is the position of the camera device when capturing the original image, and the shooting angle is the wide angle and pitch angle of the camera device when capturing the original image; The processor is used to obtain an original image captured by the camera device, and the processor is also used to process the original image according to the image restoration method according to any one of claims 1 to 3 above.

5. The image restoration system according to claim 4, wherein: It also includes a model storage module, which is used to store image restoration models that correspond one to one with meteorological categories.

6. The image restoration system according to claim 4, wherein: It also includes a parameter editing module, which is used to manually set landmark features and reference coordinates.

7. A storage module, characterized in that: The storage module stores a computer program, and when the computer program is executed, it is used to implement the image restoration method according to any one of claims 1 to 3.

Images