Image feature extraction and recognition integrated system

Abstract

The invention discloses an image feature extraction and recognition integrated system, which relates to the field of image processing and comprises a conversion module used for receiving an image to be processed and converting image data into a multi-type image data set; the setting module is used for setting the specification of a feature recognition window and configuring the feature recognition window to each image in the multi-type image data set; the analysis module is used for analyzing the feature value of the region image under the feature recognition window in the image; according to the method, after the image quality is enhanced through composite processing and the image is converted into a multi-type data set, a region image is acquired by utilizing customizable window movement, local features are calculated and captured in combination with feature values, a feature region is determined based on an analysis result, and a key region is accurately positioned through multi-dimensional fusion and iterative verification. And the feature extraction comprehensiveness and the recognition accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and specifically provides an integrated system for image feature extraction and recognition. Background Art

[0002] An image feature region refers to a local region in an image that has unique properties and can be used to distinguish or describe the image content. It is a core concept in the field of computer vision. These regions usually possess stability, and common types include corners, blobs, edges, etc. Extracting feature regions can assist a computer in tasks such as image matching, object recognition, and scene understanding, and is widely applied in fields such as security monitoring, autonomous driving, and medical image analysis.

[0003] The invention patent application with the application number 202310499435.4 discloses a feature image recognition system, which includes: a feature quantization unit for obtaining fundus color photos, establishing a mapping relationship with the morphological feature parameters of the retinal vascular structure through an encoder-decoder method, and performing feature quantization; a data dimensionality reduction unit for performing data dimensionality reduction by analyzing the structure and distribution of the quantized feature data itself to obtain the feature space with the largest difference; a prediction model establishment unit for performing spatial dimensionality reduction on the basis of the feature space using a plurality of random projection matrices respectively to obtain corresponding low-dimensional subspaces, and establishing low-dimensional feature parameter prediction models for each low-dimensional subspace respectively using a machine learning method; a model integration unit for integrating the low-dimensional feature parameter prediction models using a majority voting method to obtain an integrated model corresponding to the low-dimensional subspace; a final model unit for finally integrating the integrated model using a model stacking method to obtain a final prediction model; and a prediction unit for identifying selected features of the obtained fundus color photos according to the final prediction model. This application aims to solve the problem of "how to effectively obtain the features of fundus color photos and perform refined analysis and discrimination on them".

[0004] Based on the above, it is not difficult to see that the current image feature recognition technology is widely applied in various fields;

[0005] Therefore, we newly propose an integrated system for image feature extraction and recognition. Summary of the Invention

[0006] In view of the above-mentioned drawbacks of the prior art, the present invention provides an integrated system for image feature extraction and recognition, which can effectively solve the problems of the prior art.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions;

[0008] The present invention discloses an integrated system for image feature extraction and recognition, including:

[0009] A conversion module for receiving an image to be processed and converting the image data into a multi-type image data set; a setting module for setting the specifications of a feature recognition window and configuring the feature recognition window in each image of the multi-type image data set; an analysis module for analyzing the feature values of the regional images under the feature recognition window in the images; a selection module for obtaining the analysis results of the regional image feature values in the analysis module and selecting regional images based on the analysis results; a recognition module for recognizing whether there is an intersection area in the regional images selected by the selection module; an output module for setting the radius of the feature area, obtaining the central pixel of the intersection area recognized by the operation of the recognition module, and determining the feature area with the central pixel and the radius of the feature area;

[0010] Among them, the radius of the feature area set in the output module is the number of pixels.

[0011] Furthermore, a storage unit is provided inside the conversion module, and the storage unit is used to receive the conversion results of the conversion module for the image to be processed and store the converted various types of image data;

[0012] The conversion module converts the image data into a multi-type image data set including: grayscale image, contour image, relief image, binary image;

[0013] Before the conversion module converts the image to be processed into a multi-type image data set, it performs image quality enhancement processing on the image to be processed. After the image to be processed completes the image quality enhancement processing, it then performs the conversion operation of the multi-type image data set;

[0014] Among them, the multi-type image data set also includes the original image of the image to be processed.

[0015] Furthermore, the image quality enhancement processing formula for the image to be processed is:

[0016] ;

[0017] In the formula: is the image after image quality enhancement processing; is the structure enhancement parameter; is the Gaussian blur operator of the standard deviation of the image to be processed; is the detail sharpening parameter; ; is the global denoising parameter; represents the result after further Gaussian smoothing of the low-frequency structure layer; is the image to be processed; represents the element-by-element multiplication operation;

[0018] Among them, , , respectively represent the processing of performing structural enhancement, detail sharpening, and global denoising on the image to be processed, and the structural enhancement parameter The default value range is [0.5, 0.8], and the detail sharpening parameter The default value range is [0.3, 1], and the global denoising parameter The default value range is [0.1, 0.3].

[0019] Furthermore, a control unit is provided at a lower level of the setting module. The control unit is used to control the continuous movement of the feature recognition window on the image surface to obtain images of different regions. The control unit forwards the obtained regional images in real time to the analysis module for processing by the analysis module;

[0020] The specification of the feature recognition window is user-defined by the system end-user, and is not greater than one-fourth of the image area and not less than one-sixteenth of the image area. When the control unit controls the continuous movement of the feature recognition window on the image surface, the feature recognition window makes a straight-line movement horizontally or vertically, and there is an intersection area between two adjacent regional images obtained by the movement, and the intersection area is 1 / 4 - 1 / 2 of the area of the regional image.

[0021] Furthermore, the calculation logic of the feature value of the regional image under the feature recognition window in the grayscale image is:

[0022] ;

[0023] In the formula: is the feature response value of the pixel point (x, y); is the gradient magnitude; is the gradient direction change rate; is the local contrast;

[0024] Among them, each pixel in the regional image under the feature recognition window is calculated based on formula (1), and then summed. The summation result is recorded as the feature response value of the regional image;

[0025] The calculation logic of the feature value of the regional image under the feature recognition window in the contour image is:

[0026] ;

[0027] In the formula: is the feature representation value of the contour point (x, y); is the weight coefficient; is the curvature of the contour point; is the distance from the contour point to the concave area in the regional image; is the association strength between the contour point and the hole in the regional image;

[0028] Among them, The sum is 1, and all are non-zero positive numbers. Based on formula (2), each contour point pixel in the regional image under the feature recognition window is calculated, and then summed. The summation result is recorded as the feature representation value of the regional image.

[0029] Furthermore, the calculation logic of the feature value of the regional image under the feature recognition window in the relief image is:

[0030] ;

[0031] In the formula: is the feature score of the entire regional image; , , are the gray gradient, texture entropy, and edge density of the regional image; is the extreme value of the gray gradient in the global neighborhood of the regional image; is the extreme value of the texture entropy in the global neighborhood of the regional image; is the extreme value of the edge density in the global neighborhood of the regional image;

[0032] The calculation logic of the feature value of the regional image under the feature recognition window in the image to be processed for image quality enhancement is:

[0033] ;

[0034] In the formula: is the feature representation value of the current pixel (x, y); is the gray value of the current pixel (x, y); is the average gray value within the neighborhood N centered on (x, y); is the gray standard deviation within the neighborhood N; is a very small constant; is the gradient magnitude of the current pixel (x, y); is the global maximum value of the gradient magnitude in the image; is the gray entropy within the neighborhood N centered on (x, y); is the theoretical maximum value of the entropy value;

[0035] Among them, each pixel in the regional image is calculated based on formula (3), and then summed. The summation result is recorded as the feature representation value of the regional image.

[0036] Furthermore, the calculation logic of the feature value of the regional image under the feature recognition window in the binary image is:

[0037] ;

[0038] In the formula: is the feature reference value of the regional image; is the number of pixels with two color values in the regional image; is the total number of pixels in the regional image.

[0039] Furthermore, the result of the eigenvalue analysis of the regional image run by the analysis module is recorded as:

[0040] ;

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] Five sets of eigenvalue calculation results are obtained, and the calculation results in each of the five sets are arranged in descending order;

[0046] Obtain the regional images corresponding to the first calculation results in the five sets, and then the recognition module identifies whether there is an intersection among the regional images. If there is, jump to the output module to run;

[0047] If not, obtain the regional images that have no intersection with other regional images, use the obtained regional images as the iteration target, and perform iteration with the regional images corresponding to the eigenvalues adjacent to the eigenvalues in the eigenvalue calculation result set corresponding to the iteration target.

[0048] Furthermore, a jump unit is provided at the lower levels of the selection module and the recognition module. The jump unit is triggered when the recognition result of the recognition module is no, and jumps to the selection module to run again until the recognition result is yes after the recognition module runs again following the selection module, and then it ends.

[0049] Furthermore, a storage unit is internally connected to the conversion module through wireless network interaction. The conversion module is connected to the setting module through wireless network interaction. The lower level of the setting module is connected to the control unit through wireless network interaction. The setting module is connected to the analysis module and the selection module through wireless network interaction. The analysis module is connected to the control unit through wireless network interaction. The selection module is connected to the recognition module and the output module through wireless network interaction. The selection module and the lower level of the recognition module are connected to the jump unit through wireless network interaction.

[0050] Adopting the technical solution provided by the present invention, compared with the known prior art, it has the following beneficial effects:

[0051] The present invention provides an integrated system for image feature extraction and recognition. During the operation of this system, first, image quality enhancement is performed on the image. Through a composite process combining structural enhancement, detail sharpening, and global denoising, the image quality is effectively improved, laying a foundation for subsequent analysis. Then, the image is converted into a multi-type dataset including grayscale, contour, relief, binary images, and the original image, and analyzed from different feature dimensions. By continuously moving a feature recognition window with customizable specifications on the image and obtaining partially overlapping regional images, combined with the feature value calculation logic corresponding to each type of image, accurate capture of local features of the image is achieved. Furthermore, based on the analysis results of the feature values, regional images are selected, and the feature region is determined by judging whether there is an intersection region. This multi-dimensional fusion analysis and iterative verification mechanism can accurately locate the key feature region of the image, effectively improving the comprehensiveness of feature extraction and the accuracy of recognition, and seeking a new image feature recognition and extraction solution for the field of image analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0053] Figure 1 It is a schematic structural diagram of an integrated system for image feature extraction and recognition. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0055] The following further describes the present invention with reference to the embodiments.

[0056] Embodiment:

[0057] An integrated system for image feature extraction and recognition in this embodiment, as Figure 1 shown, includes:

[0058] A conversion module for receiving the image to be processed and converting the image data into a multi-type image dataset;

[0059] The conversion module is internally provided with a storage unit, which is used to receive the conversion results of the conversion module for the image to be processed and store various types of image data obtained by the conversion;

[0060] The conversion module converts the image data into a multi-type image data set, including: grayscale image, contour image, relief image, binary image;

[0061] Before converting the image to be processed into a multi-type image data set, the conversion module performs image quality enhancement processing on the image to be processed. After the image to be processed completes the image quality enhancement processing, the conversion operation of the multi-type image data set is then executed;

[0062] Among them, the multi-type image data set also includes the original image of the image to be processed;

[0063] The image quality enhancement processing formula for the image to be processed is:

[0064] ;

[0065] In the formula: is the image after image quality enhancement processing; is the structure enhancement parameter; is the Gaussian blur operator of the standard deviation of the image to be processed; is the detail sharpening parameter; ; is the global denoising parameter; represents the result after further Gaussian smoothing of the low-frequency structure layer; is the image to be processed; represents the operation of element-wise multiplication;

[0066] Among them, , , respectively represent the processing of structure enhancement, detail sharpening and global denoising on the image to be processed. The structure enhancement parameter has a default value range of [0.5, 0.8], the detail sharpening parameter has a default value range of [0.3, 1], and the global denoising parameter has a default value range of [0.1, 0.3];

[0067] Through the above logical formula calculation, the image to be processed is enhanced to ensure that the subsequent processing results of the system for the image to be processed are more accurate;

[0068] The setting module is used to set the feature recognition window specifications and configure the feature recognition window for each image in the multi-type image data set;

[0069] The lower-level setting of the setting module has a control unit, which is used to control the continuous movement of the feature recognition window on the image surface to obtain images of different regions. The control unit forwards the obtained regional images to the analysis module in real time for processing by the analysis module;

[0070] The specification of the feature recognition window is user-defined by the system-side user, and is not greater than one-fourth of the image area and not less than one-sixteenth of the image area. When the control unit controls the continuous movement of the feature recognition window on the image surface, the feature recognition window moves in a straight line horizontally or vertically, and there is an intersection area between two adjacent regional images obtained by the movement, and the intersection area is 1 / 4 to 1 / 2 of the area of the regional image;

[0071] An analysis module, which is used to analyze the feature values of the regional image under the feature recognition window in the image;

[0072] The calculation logic of the feature value of the regional image under the feature recognition window in the grayscale image is:

[0073] ;

[0074] In the formula: is the feature response value of the pixel point (x, y); is the gradient amplitude; is the gradient direction change rate; is the local contrast;

[0075] Among them, each pixel in the regional image under the feature recognition window is calculated based on formula (1), and then summed up. The summation result is recorded as the feature response value of the regional image;

[0076] The calculation logic of the feature value of the regional image under the feature recognition window in the contour image is:

[0077] ;

[0078] In the formula: is the feature representation value of the contour point (x, y); is the weight coefficient; is the contour point curvature; is the distance from the contour point to the concave area in the regional image; is the association strength between the contour point and the hole in the regional image;

[0079] Among them, The sum of is 1, and they are all non-zero positive numbers. Each contour point pixel in the regional image under the feature recognition window is calculated based on formula (2), and then summed up. The summation result is recorded as the feature representation value of the regional image;

[0080] The calculation logic of the feature value of the regional image under the feature recognition window in the embossed image is:

[0081] ;

[0082] In the formula: is the global feature score of the regional image; , , are the gray gradient, texture entropy, and edge density of the regional image; is the extreme value of the gray gradient in the global neighborhood of the regional image; is the extreme value of the texture entropy in the global neighborhood of the regional image; is the extreme value of the edge density in the global neighborhood of the regional image;

[0083] The calculation logic of the feature value of the regional image under the feature recognition window in the image quality enhancement process is as follows:

[0084] ;

[0085] In the formula: is the feature representation value of the current pixel (x, y); is the gray value of the current pixel (x, y); is the average gray value in the neighborhood N centered on (x, y); is the gray standard deviation in the neighborhood N; is a very small constant; is the gradient magnitude of the current pixel (x, y); is the global maximum value of the gradient magnitude in the image; is the gray entropy in the neighborhood N centered on (x, y); is the theoretical maximum value of the entropy value;

[0086] Among them, each pixel in the regional image is calculated based on formula (3), and then summed up. The summation result is recorded as the feature representation value of the regional image;

[0087] The calculation logic of the feature value of the regional image under the feature recognition window in the binary image is as follows:

[0088] ;

[0089] In the formula: is the feature reference value of the regional image; is the number of pixels with two color values in the regional image; is the total number of pixels in the regional image;

[0090] By using the above logical formula, the eigenvalue calculation is performed on the regional images of each type of image under the feature recognition window, providing operating data support for the selection module and recognition module in the system of this embodiment, and ensuring that the system in this embodiment can stably identify and extract the feature regions in the image to be processed;

[0091] The result of analyzing the eigenvalue of the regional image by the analysis module is denoted as:

[0092] ;

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] Five sets of eigenvalue calculation results are obtained, and the calculation results in each of the five sets are arranged in descending order;

[0098] Obtain the regional images corresponding to the first calculation results in the five sets, and then the recognition module identifies whether there is an intersection in each regional image. If there is, jump to the output module to run;

[0099] If not, obtain the regional images that have no intersection with other regional images, use the obtained regional images as the iteration target, and perform iteration with the regional images corresponding to the eigenvalues sorted adjacent to the eigenvalues in the eigenvalue calculation result set corresponding to the iteration target;

[0100] The selection module is used to obtain the analysis results of the eigenvalues of the regional images in the analysis module and select regional images based on the analysis results;

[0101] The recognition module is used to identify whether there is an intersection region in the regional images selected by the selection module;

[0102] A jump unit is set under the selection module and the recognition module. The jump unit is triggered when the recognition result of the recognition module is negative, and jumps to the selection module to run again until the recognition result is positive after the recognition module runs again following the selection module, and then it ends;

[0103] The output module is used to set the radius of the feature region, obtain the central pixel of the intersection region recognized by the recognition module running, and determine the feature region based on the central pixel and the radius of the feature region;

[0104] Among them, the radius of the feature region set in the output module is the number of pixels;

[0105] Inside the conversion module, there is a storage unit connected through wireless network interaction. The conversion module is interactively connected with the setting module through wireless network. The lower level of the setting module is interactively connected with the control unit through wireless network. The setting module is interactively connected with the analysis module and the selection module through wireless network. The analysis module is interactively connected with the control unit through wireless network. The selection module is interactively connected with an identification module and an output module through wireless network. The selection module and the lower level of the identification module are interactively connected with a jump unit through wireless network.

[0106] In this embodiment, the conversion module runs to receive the image to be processed, converts the image data into a multi-type image data set. The storage unit synchronously receives the conversion result of the image to be processed by the operation of the conversion module, and stores each type of converted image data. The setting module runs later to set the specification of the feature recognition window, and configures the feature recognition window on each image in the multi-type image data set. The control unit synchronously controls the continuous movement of the feature recognition window on the image surface to obtain images of different regions. The control unit forwards the obtained regional images in real time to the analysis module for processing by the analysis module. Then, the analysis module analyzes the feature values of the regional images under the feature recognition window in the image. The selection module further obtains the analysis result of the regional image feature values in the analysis module, selects the regional image based on the analysis result. The identification module identifies whether there is an intersection region in the regional image selected by the selection module. The jump unit is triggered when the identification result of the identification module is no, and jumps to the selection module to run again until after the identification module runs again following the selection module, when the identification result is yes, it ends. Finally, the output module sets the radius of the feature region, obtains the central pixel of the intersection region recognized by the operation of the identification module, and determines the feature region based on the central pixel and the radius of the feature region.

[0107] Through the operation of the system in the above embodiment, a brand-new feature recognition and extraction solution for images adapting to a wide range of scenarios is provided, ensuring that the application of image feature recognition and extraction technology is more universal.

[0108] In summary, during the operation of the system in the above embodiment, the image quality is enhanced for the image, and a composite process combining structure enhancement, detail sharpening, and global denoising is performed, effectively improving the image quality and laying a foundation for subsequent analysis. The image is converted into a multi-type data set including grayscale, contour, relief, binary image, and the original image, and analyzed from different feature dimensions. The feature recognition window with customizable specifications continuously moves on the image and obtains partially overlapping regional images. Combining the feature value calculation logics corresponding to each type of image, the accurate capture of local features of the image is realized. The regional image is selected based on the analysis result of the feature values, and the feature region is determined by judging whether there is an intersection region. This multi-dimensional fusion analysis and iterative verification mechanism can accurately locate the key feature regions of the image, effectively improving the comprehensiveness of feature extraction and the accuracy of recognition, and seeking a brand-new image feature recognition and extraction solution for the field of image analysis.

[0109] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements will not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An integrated system for image feature extraction and recognition, characterized in that, Including: A conversion module, configured to receive an image to be processed and convert the image data into a multi-type image data set; A setting module, configured to set the specification of a feature recognition window and configure the feature recognition window for each image in the multi-type image data set; An analysis module, configured to analyze the feature values of the area image under the feature recognition window in the image; A selection module, configured to obtain the analysis result of the area image feature values in the analysis module and select the area image based on the analysis result; A recognition module, configured to recognize whether there is an intersection area in the area image selected by the selection module; An output module, configured to set the radius of the feature area, obtain the central pixel of the intersection area recognized by the recognition module during operation, and determine the feature area based on the central pixel and the radius of the feature area; Wherein, the radius of the feature area set in the output module is the number of pixels.

2. The integrated system for image feature extraction and recognition according to claim 1, wherein A storage unit is arranged inside the conversion module, and the storage unit is configured to receive the conversion result of the image to be processed by the operation of the conversion module and store each type of image data obtained by conversion; The conversion module converts the image data into a multi-type image data set including: grayscale image, contour image, relief image, binary image; Before the conversion module converts the image to be processed into a multi-type image data set, it performs image quality enhancement processing on the image to be processed. After the image quality enhancement processing of the image to be processed is completed, the conversion operation of the multi-type image data set is then executed; Wherein, the multi-type image data set further includes the original image of the image to be processed.

3. An integrated system for image feature extraction and recognition according to claim 2, wherein The image quality enhancement processing formula of the image to be processed is: ； Where: is the image after image quality enhancement processing; is the structure enhancement parameter; is the Gaussian blur operator of the standard deviation of the image to be processed; is the detail sharpening parameter; ; is the global denoising parameter; represents the result after further Gaussian smoothing of the low-frequency structure layer; is the image to be processed; represents the operation of element-wise multiplication; Among them, , , respectively represent the processing of structural enhancement, detail sharpening, and global denoising on the image to be processed. The structural enhancement parameter has a default value range of [0.5, 0.8]. The detail sharpening parameter has a default value range of [0.3, 1]. The global denoising parameter has a default value range of [0.1, 0.3].

4. An integrated system for image feature extraction and recognition according to claim 1, characterized in that, A control unit is arranged at a lower level of the setting module, and the control unit is configured to control the continuous movement of the feature recognition window on the image surface to obtain area images of different regions. The area images obtained by the operation of the control unit are forwarded to the analysis module in real time for processing by the analysis module; The specification of the feature recognition window is user-defined by the system end user, and is not greater than one-fourth of the image area and not less than one-sixteenth of the image area. When the control unit controls the continuous movement of the feature recognition window on the image surface, the feature recognition window moves in a straight line horizontally or vertically, and there is an intersection area between two adjacent area images obtained by the movement, and the intersection area is 1 / 4 to 1 / 2 of the area of the area image.

5. An integrated system for image feature extraction and recognition according to claim 1, characterized in that, The calculation logic of the feature values of the area image under the feature recognition window in the grayscale image is: ； In the formula: is the feature response value of the pixel point (x, y); is the gradient magnitude; is the gradient direction change rate; is the local contrast; Wherein, each pixel in the area image under the feature recognition window is calculated based on formula (1), and then summed, and the summation result is recorded as the feature response value of the area image; The calculation logic of the feature values of the area image under the feature recognition window in the contour image is: ； Wherein: is the characteristic value of the contour point (x, y); is the weight coefficient; is the contour point curvature; is the distance from the contour point to the concave region in the regional image; is the association strength between the contour point and the hole in the regional image; wherein, the sum is 1, and all are non-zero positive numbers. Based on formula (2), each contour point pixel in the area image under the feature recognition window is calculated and then summed up, and the summation result is recorded as the feature representation value of the area image.

6. The integrated system for image feature extraction and recognition according to claim 1, characterized in that, The calculation logic of the feature values of the area image under the feature recognition window in the relief image is: ； Wherein: is the feature score of the entire regional image; , , are the gray gradient, texture entropy, and edge density of the regional image; is the extreme value of the gray gradient in the global neighborhood of the regional image; is the extreme value of the texture entropy in the global neighborhood of the regional image; is the extreme value of the edge density in the global neighborhood of the regional image; The calculation logic of the feature values of the area image under the feature recognition window in the image to be processed after image quality enhancement processing is: ； Where: is the characteristic representation value of the current pixel (x, y); is the gray value of the current pixel (x, y); is the average gray value in the neighborhood N centered at (x, y); is the grayscale standard deviation within the neighborhood N; is a very small constant; is the gradient magnitude of the current pixel (x, y); is the global maximum value of the gradient amplitude in the image; is the grayscale entropy within the neighborhood N centered at (x, y); is the theoretical maximum value of entropy; Wherein, each pixel in the area image is calculated based on formula (3), and then summed, and the summation result is recorded as the feature representation value of the area image.

7. An integrated system for image feature extraction and recognition according to claim 1, characterized in that, The calculation logic of the feature values of the area image under the feature recognition window in the binary image is: ； Wherein: is the characteristic reference value of the regional image; is the number of pixels of two color values in the regional image; is the total amount of pixels in the regional image.

8. An integrated system for image feature extraction and recognition according to claim 1, characterized in that, The result of the analysis module running and analyzing the feature values of the area image is recorded as: ； ； ； ； ； Five sets of feature value calculation results are obtained, and the calculation results in the five sets are arranged in descending order; Obtain the regional images corresponding to the first calculation results in the five sets, and then the recognition module identifies whether there is an intersection among the regional images. If there is, jump to the output module to run; If not, obtain the regional images that have no intersection with other regional images, use the obtained regional images as the iteration target, and iterate with the regional images corresponding to the eigenvalues sorted adjacent to each other in the set of eigenvalue calculation results corresponding to the iteration target.

9. An integrated system for image feature extraction and recognition according to claim 1, characterized in that, A jump unit is set under the selection module and the recognition module. The jump unit is triggered when the recognition result of the recognition module is no, and jumps to the selection module to run again until the recognition result is yes after the recognition module runs again following the selection module, and then ends.

10. An integrated system for image feature extraction and recognition according to claim 1, characterized in that, Inside the conversion module, there is a storage unit connected through wireless network interaction. The conversion module is interactively connected with the setting module through wireless network. Under the setting module, it is interactively connected with the control unit through wireless network. The setting module is interactively connected with the analysis module and the selection module through wireless network. The analysis module is interactively connected with the control unit through wireless network. The selection module is interactively connected with the recognition module and the output module through wireless network. The selection module and the recognition module are interactively connected with the jump unit through wireless network.

Images