An intelligent iris recognition system for canines

Abstract

The present invention relates to the field of pet feature recognition technology, and in particular to a canine intelligent iris recognition system, comprising: a sample image acquisition module for acquiring sample iris images, a sample feature extraction module for extracting canine iris features and storing the canine iris features in a server, a target image acquisition module for acquiring shooting light intensity and acquiring a target iris image according to the shooting light intensity, an acquisition screening module for adjusting the acquisition process of the target iris image, a target feature extraction module for extracting target canine iris features according to the target iris image, a feature matching module for performing iris feature matching between the target canine iris feature extraction result and the canine iris feature extraction result, and an iterative matching module for iterating the iris feature matching process. The present invention effectively improves the efficiency and accuracy of canine iris recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of pet feature recognition, and in particular to an intelligent iris recognition system for canine animals. Background Art

[0002] With the development of science and technology, biometric recognition technology has become one of the important means of identity verification. Currently, face recognition, fingerprint recognition and other technologies on the market have been widely used in human identity verification. However, the identification of pets such as dogs is still in its infancy. The iris of canine animals has unique texture features, and the uniqueness and stability of this feature make it an ideal biometric recognition target.

[0003] Chinese patent publication number CN106096526A discloses an iris recognition method and iris recognition system, which are used to solve the technical problems of large amount of data transmitted and heavy burden of encryption and decryption when performing iris recognition. The method includes: an image processing unit obtains a first image including facial information of a first user; the image processing unit intercepts a second image including binocular image information of the first user from the first image; wherein the data amount of the second image is less than the data amount of the first image; the image processing unit performs a first encryption process on the second image to obtain an encrypted second image; the image processing unit sends the encrypted second image to the iris recognition unit; it can be seen that the invention only analyzes human iris features and faces, and does not distinguish and analyze irises of different iris shapes and different iris patterns, and there are problems of low efficiency and low accuracy in iris recognition. Summary of the invention

[0004] The object of the present invention is to provide an intelligent iris recognition system for canines to solve at least one of the problems existing in the prior art.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A sample image acquisition module, used for acquiring a sample iris image;

[0007] A sample feature extraction module, used to divide the sample iris image into regions, and extract edge features, texture features and color features according to the region division results, and use the extraction results as canine iris features. The sample feature extraction module is also used to store the canine iris features in a server;

[0008] A target image acquisition module is used to acquire a target iris image that meets the set standards;

[0009] An acquisition and screening module is used to analyze the saturation and contrast of the target iris image, and adjust the acquisition process of the target iris image according to the analysis results of the saturation and contrast of the target iris image;

[0010] A target feature extraction module is used to extract the iris features of the target dog according to the target iris image;

[0011] A feature matching module, used for performing iris feature matching between the target dog iris feature extraction result and the dog iris feature extraction result;

[0012] The iterative matching module is used to iterate the process of iris feature matching and send the iterative result of iris feature matching to the user.

[0013] Furthermore, the sample feature extraction module is provided with a region division unit, and the region division unit is used to divide the sample iris image into regions;

[0014] The region division unit clusters the pixels of the iris image according to the grayscale values ​​g(x,y) of the pixels of the sample iris image to obtain clusters;

[0015] Where g(x,y) is the gray value of the pixel whose horizontal coordinate is x and whose vertical coordinate is y;

[0016] The region division unit divides the pixels included in each cluster into continuous regions according to the coordinates of the pixels included in each cluster: the region division unit combines the pixels included in the same continuous cluster into divided regions;

[0017] The region division unit classifies each divided region according to the average gray value Gray(i) of each divided region: if Gray(i)∈{[0,5],[0,5],[0,5]}, the region division unit determines that the divided region is the sample pupil region; if Gray(i)∈{[200,255],[240,255],[240,255]}, the region division unit determines that the divided region is the sample sclera region; if g(x,y)∈{[0,5],[0,5],[0,5]} and g(x,y) {[200,255],[240,255],[240,255]}, the region division unit takes the pixel point as the sample iris pixel point, and combines the sample iris pixel points into a sample iris region; the region division unit determines the region composed of the remaining pixel points as the remaining region.

[0018] Furthermore, the sample feature extraction module is further provided with an edge feature extraction unit, and the edge feature extraction unit analyzes the edge feature according to the region division result of the sample iris image;

[0019] The edge feature extraction unit uses the intersection pixel points of the sample iris area and the sample pupil area as the inner iris edge curve pixel points, and uses the intersection pixel points of the sample sclera area and the sample iris area as the outer iris edge pixel points;

[0020] The edge feature extraction unit connects the inner iris pixel points into an inner iris curve and connects the outer iris pixel points into an outer iris curve;

[0021] The edge feature extraction unit judges the sample iris image according to the analysis result of the inner iris curve: if the inner iris curve is a circular curve, the edge feature extraction unit judges that the sample iris image is qualified; if the inner iris curve is a non-circular curve, the edge feature extraction unit judges that the sample iris image is unqualified;

[0022] The edge feature extraction unit stores the outer iris curve as an edge feature.

[0023] Furthermore, the sample feature extraction module is further provided with a texture feature extraction unit, which extracts texture features of the sample iris according to the judgment result and the area division result of the sample iris image;

[0024] The texture feature extraction unit performs grayscale processing on the sample iris area: the combing feature analysis unit sets the grayscale value of the pixel points in the sample iris area to G(x,y), and sets G(x,y)=|g(x,y)|;

[0025] The texture feature extraction unit clusters the pixel points of the sample iris area according to the grayscale values ​​of the pixel points to obtain various texture clusters;

[0026] The texture feature analysis unit connects edge curves s(j) of each texture cluster according to the pixel points in each texture cluster;

[0027] The texture feature extraction unit judges the sample iris pattern feature according to the edge curve s(j) of each texture cluster, and sets the recognition mode index according to the judgment result of the sample iris shape feature: if the sample iris pattern feature is radial, the texture feature extraction unit sets the recognition mode index to A1, sets A1=1, and uses the radial starting point pixel coordinate q1(m) and the radial end point pixel coordinate q2(m) as sample texture data; if the sample iris pattern feature is mesh, the texture feature extraction unit sets the recognition mode to A2, sets A2=2, and uses the mesh node wz(f) as sample texture data; if the sample iris pattern feature is other, the texture feature extraction unit sets the recognition mode to A3, sets A3=3, and uses the edge curve s(j) of each texture cluster as sample texture data;

[0028] Wherein, q1(m) is the starting pixel coordinate of the m-th radial line, q2(m) is the ending pixel coordinate of the m-th radial line, wz(f) is the pixel coordinate of the f-th mesh node, and m and f are both digital subscripts;

[0029] The texture feature extraction unit uses the recognition pattern index Ah and the sample texture data as the texture feature extraction result, and sets h=1, 2, 3.

[0030] Furthermore, the sample feature extraction module is further provided with a color feature extraction unit, and the color feature extraction unit is used to extract the sample color feature according to the area division result of the sample iris image;

[0031] The color feature extraction unit uses the average gray value Gray(i) of the sample pupil area as the first sample color feature C(1);

[0032] The color feature extraction unit uses the average grayscale value WL(j) of each texture cluster of the sample iris region as the second sample color feature C(2);

[0033] The color feature extraction unit uses the first color feature, the second color feature, and the texture feature extraction results as feature extraction results of the sample iris image.

[0034] Furthermore, the target image acquisition module compares the shooting light intensity w with the preset shooting light intensity W, and acquires the target iris image according to the comparison result: if W1≤w<W2, the target image acquisition module determines that the lighting environment is normal, and acquires the target iris image; if w≥W2 or w<W1, the target image acquisition module determines that the lighting environment is abnormal, and does not acquire the target iris image;

[0035] Wherein, W1 is the first preset light intensity, W2 is the second preset light intensity, and W1<W2.

[0036] Furthermore, the acquisition and screening module is provided with a saturation analysis unit, and the saturation analysis unit is used to analyze the saturation of the target iris image and adjust the acquisition process of the target iris image according to the saturation analysis result;

[0037] The saturation analysis unit converts the target iris image into an HSV color image and calculates the target saturation index η, setting ;

[0038] Among them, max(R,G,B) represents the maximum value of the three color channels (R,G,B) in the HSV color image, min(R,G,B) represents the minimum value of the three color channels (R,G,B) in the HSV color image, and V is the brightness of the HSV color image;

[0039] The saturation analysis unit compares the target saturation index η with the saturation index threshold N, and adjusts the acquisition process of the target iris image according to the comparison result: if η≥N, the saturation analysis unit determines that the saturation is high, and adjusts the second preset light intensity to W2'; if η<N, the saturation analysis unit determines that the saturation is normal and does not make any adjustment;

[0040] The acquisition and screening module is also provided with a contrast analysis unit, which is used to analyze the contrast of the target iris image and optimize the acquisition and adjustment process of the target iris image according to the saturation analysis result;

[0041] The contrast analysis unit calculates the contrast C of the target iris image and sets , the contrast analysis unit optimizes the acquisition and adjustment process of the target iris image according to the contrast C, and optimizes the saturation index threshold N to N';

[0042] Wherein, T is the number of pixels of the target iris image, gmr(i) is the i-th pixel of the target iris image, and Pgmr is the mean grayscale value of the pixels of the target iris image.

[0043] Further, the feature matching module is provided with an edge feature matching unit, which is used to match the target edge feature with the edge feature;

[0044] The edge feature matching unit calculates the edge similarity α according to the outer iris curve and the target outer iris curve, and sets α=Σ[(az-cz) 2 +(bz-dz) 2 ] 1 / 2 / K; where z is a mathematical subscript, set z=1,2...K, K is the number of pixels of the outer iris curve; az is the horizontal coordinate of the zth pixel in the target edge curve, cz is the horizontal coordinate of the zth pixel in the edge curve, bz is the vertical coordinate of the zth pixel in the target edge curve, and dz is the vertical coordinate of the zth pixel in the edge curve;

[0045] The edge feature matching unit compares the edge similarity α with the edge similarity threshold B, and analyzes the edge feature matching result according to the comparison result: if α<B, the edge feature matching unit determines that the edge feature matching is successful; if α≥B, the edge feature matching unit determines that the edge feature matching fails.

[0046] Further, the texture feature matching unit is used to perform texture feature matching on the target texture feature and the texture feature: when Ah≠ASh, the texture feature matching unit determines that the texture features do not match; when Ah=ASh=1, the texture feature matching unit calculates a first offset index γ1, and sets γ1=[q1(m)-q2(m)]·[qs1(m)-qs2(m)]; when Ah=ASh=2, the texture feature matching unit calculates a second offset index γ2, and sets γ2=Σ|wz(f)-wzs(f)| / F; when Ah=ASh=3, the texture feature matching unit calculates a third offset index γ3;

[0047] Among them, ASh represents the index of each target recognition mode, including AS1, AS2, AS3, and AS4;

[0048] The texture feature matching unit compares each offset index γu with a preset offset index Y, sets u=1, 2, 3, and analyzes the texture matching result according to the comparison result: if γu≥Y, the texture feature matching unit determines that the texture matching fails; if γu<Y, the texture feature matching unit determines that the texture matching succeeds;

[0049] The feature matching module is further provided with a color feature matching unit, which is used to perform color feature matching on the target color feature and the color feature: when C(1)=CS(1), the color feature matching unit determines that the first color feature is matched successfully. At this time, if C(2)=CS(2), the color feature matching unit determines that the color matching is successful; if C(2)≠CS(2), the color feature matching unit determines that the color matching fails; when C(1)≠CS(1), the color feature matching unit determines that the color feature matching fails;

[0050] Among them, CS(2) is the target second color feature, and CS(1) is the target first color feature.

[0051] Furthermore, the iterative matching module is used to iterate the iris feature matching process;

[0052] The iterative matching module iterates the iris feature matching process according to the canine iris features stored in the server: when there are still unmatched canine iris features in the server, if the iris feature matching result of the current canine iris feature is a failure, the iterative matching module uses the next canine iris feature as the canine iris feature extraction result of the next iris matching process; if the iris feature matching result of the current canine iris feature is a success, the iterative matching module stops the iterative process and sends the iris feature matching result to the user; when there are no unmatched canine iris features in the server, if the iris feature matching result of the current canine iris feature is a failure, the iterative matching module sends a "matching failure" to the user; if the iris feature matching result of the current canine iris feature is a success, the iterative matching module stops the iterative process and sends the iris feature matching result to the user. Further,

[0053] Compared with the prior art, the beneficial effects of the present invention are that the sample image acquisition module and the target image acquisition module are used to acquire sample iris images and target iris images, thereby improving the accuracy and completeness of image acquisition; the sample feature extraction module is used to extract canine iris features, thereby improving the accuracy of feature extraction; the acquisition screening module is used to adjust the acquisition process of the target iris image, thereby improving the accuracy and completeness of image acquisition; the target feature extraction module is used to extract the target canine iris features, thereby improving the efficiency and accuracy of feature extraction; the feature matching module is used to match single features, thereby improving the efficiency of feature matching; the iterative matching module is used to match the canine iris features stored in the server one by one, thereby improving the efficiency and accuracy of canine iris recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0055] Figure 1 Schematic diagram of the structure of the canine intelligent iris recognition system of this embodiment.

[0056] Figure 2 Schematic diagram of the structure of the sample feature extraction module in this embodiment.

[0057] Figure 3 This is a schematic diagram of the structure of the collection and screening module in this embodiment.

[0058] Figure 4 Schematic diagram of the structure of the feature matching module of this embodiment. DETAILED DESCRIPTION

[0059] In order to more clearly illustrate the present invention, the present invention is further described below in conjunction with preferred embodiments and accompanying drawings. Similar components in the accompanying drawings are represented by the same reference numerals. It should be understood by those skilled in the art that the content specifically described below is illustrative rather than restrictive, and should not be used to limit the scope of protection of the present invention.

[0060] It should be noted that, although the terms first, second, third, etc. may be used to describe in the embodiments of the present application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, without departing from the scope of the embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first.

[0061] Specifically, the intelligent iris recognition system for canine animals described in this embodiment is applied to iris recognition of outdoor canine pets; the intelligent iris recognition system for canine animals acts on a small device that can perform iris recognition of pets outdoors.

[0062] See also Figure 1 As shown, it is a schematic diagram of the structure of the canine intelligent iris recognition system of this embodiment, including:

[0063] The sample image acquisition module is used to acquire sample iris images.

[0064] Specifically, the sample iris image is an iris image of a canine animal in RGB image format for iris recognition. This embodiment does not specifically limit the method for obtaining the sample iris image, and those skilled in the art can freely set it as long as the requirements for obtaining the sample iris image are met.

[0065] Please continue reading Figure 1 As shown, the intelligent iris recognition system for canines also includes a sample feature extraction module, which is used to divide the sample iris image into regions, and extract edge features, texture features and color features according to the region division results, and use the extraction results as canine iris features. The sample feature extraction module is also used to store the canine iris features in a server, and the sample feature extraction module is connected to the sample image acquisition module.

[0066] See also Figure 2 As shown, the sample feature extraction module includes a region division unit, which is used to divide the sample iris image into regions;

[0067] The region division unit clusters the pixels of the iris image according to the grayscale values ​​g(x,y) of the pixels of the sample iris image to obtain clusters;

[0068] Where g(x,y) is the gray value of the pixel whose horizontal coordinate is x and whose vertical coordinate is y;

[0069] The region division unit divides the pixels included in each cluster into continuous regions according to the coordinates of the pixels included in each cluster: the region division unit combines the pixels included in the same continuous cluster into divided regions;

[0070] The region division unit classifies each divided region according to the average gray value Gray(i) of each divided region: if Gray(i)∈{[0,5],[0,5],[0,5]}, the region division unit determines that the divided region is the sample pupil region; if Gray(i)∈{[200,255],[240,255],[240,255]}, the region division unit determines that the divided region is the sample sclera region; if g(x,y)∈{[0,5],[0,5],[0,5]} and g(x,y) {[200,255],[240,255],[240,255]}, the region division unit takes the pixel point as the sample iris pixel point, and combines the sample iris pixel points into a sample iris region; the region division unit determines the region composed of the remaining pixel points as the remaining region. The region division unit divides the sample iris image to accurately segment the iris region, thereby improving the accuracy of each feature extraction.

[0071] Specifically, the specific format of the grayscale value in this embodiment is (R, G, B), that is, g(x, y) = (R, G, B), R is the red grayscale value identifier, G is the green grayscale value identifier, B is the blue grayscale value identifier, and R, G and B are all between the interval [0, 255].

[0072] Please continue reading Figure 2 As shown, the sample feature extraction module is further provided with an edge feature extraction unit, and the edge feature extraction unit analyzes the edge feature according to the region division result of the sample iris image;

[0073] The edge feature extraction unit uses the intersection pixel points of the sample iris area and the sample pupil area as the inner iris edge curve pixel points, and uses the intersection pixel points of the sample sclera area and the sample iris area as the outer iris edge pixel points;

[0074] The edge feature extraction unit connects the inner iris pixel points into an inner iris curve and connects the outer iris pixel points into an outer iris curve;

[0075] The edge feature extraction unit judges the sample iris image according to the analysis result of the inner iris curve: if the inner iris curve is a circular curve, the edge feature extraction unit judges that the sample iris image is qualified; if the inner iris curve is a non-circular curve, the edge feature extraction unit judges that the sample iris image is unqualified; the edge feature extraction unit screens out non-circular pupils by judging the inner iris curve, and extracts the outer iris curve as an edge feature to identify iris shapes such as almond and circular;

[0076] The edge feature extraction unit stores the outer iris curve as an edge feature.

[0077] Please continue reading Figure 2 As shown, the sample feature extraction module further includes a texture feature extraction unit, which extracts texture features of the sample iris according to the judgment result and the area division result of the sample iris image;

[0078] The texture feature extraction unit performs grayscale processing on the sample iris area: the combing feature analysis unit sets the grayscale value of the pixel points in the sample iris area to G(x,y), and sets G(x,y)=|g(x,y)|;

[0079] The texture feature extraction unit clusters the pixel points of the sample iris area according to the grayscale values ​​of the pixel points to obtain various texture clusters;

[0080] The texture feature analysis unit connects edge curves s(j) of each texture cluster according to the pixel points in each texture cluster;

[0081] The texture feature extraction unit judges the sample iris pattern feature according to the edge curve s(j) of each texture cluster, and sets the recognition mode index according to the judgment result of the sample iris shape feature: if the sample iris pattern feature is radial, the texture feature extraction unit sets the recognition mode index to A1, sets A1=1, and uses the radial starting point pixel coordinate q1(m) and the radial end point pixel coordinate q2(m) as sample texture data; if the sample iris pattern feature is mesh, the texture feature extraction unit sets the recognition mode to A2, sets A2=2, and uses the mesh node wz(f) as sample texture data; if the sample iris pattern feature is other, the texture feature extraction unit sets the recognition mode to A3, sets A3=3, and uses the edge curve s(j) of each texture cluster as sample texture data;

[0082] Wherein, q1(m) is the starting pixel coordinate of the mth radial line, q2(m) is the ending pixel coordinate of the mth radial line, wz(f) is the pixel coordinate of the fth mesh node, and m and f are both digital subscripts; the texture feature extraction unit improves the efficiency of subsequent texture feature matching by dividing the iris texture pattern into radial, mesh and others;

[0083] The texture feature extraction unit uses the recognition pattern index Ah and the sample texture data as the texture feature extraction result, and sets h=1, 2, 3.

[0084] Specifically, the data format of the grayscale value of the pixel after the grayscale processing is a single numerical value. In this embodiment, |g(x,y)| represents the grayscale processing of the color image, and its processing formula is G(x,y)=|g(x,y)|=0.299R+0.578G+0.114B; at the same time, the process of "connecting the edge curves s(j) of each texture clustering cluster according to the pixel points in each texture clustering cluster" in this embodiment is the same as the process of analyzing the inner iris curve and the outer iris curve mentioned above, and this embodiment will not be repeated here.

[0085] Specifically, the "sample iris pattern features" described in this embodiment include radial, mesh and others; this embodiment specifically limits the judgment process of the sample iris pattern features based on the existing technology and common knowledge, and those skilled in the art can freely set it as long as the requirements of the judgment process of the sample iris pattern features are met. For example, in this embodiment, radial and mesh sample images can be input for pre-training through machine learning to obtain a training model, and the sample iris image is input into the obtained training model to judge the sample iris pattern features.

[0086] Please continue reading Figure 2 As shown, the sample feature extraction module further includes a color feature extraction unit, and the color feature extraction unit is used to extract the sample color feature according to the area division result of the sample iris image;

[0087] The color feature extraction unit uses the average gray value Gray(i) of the sample pupil area as the first sample color feature C(1);

[0088] The color feature extraction unit uses the average grayscale value WL(j) of each texture cluster of the sample iris region as the second sample color feature C(2);

[0089] The color feature extraction unit uses the first color feature, the second color feature, and the texture feature extraction results as the feature extraction results of the sample iris image; the color feature extraction unit sets the first sample color feature and the second sample color feature to perform two levels of features on the color setting to improve the efficiency of color feature matching.

[0090] Specifically, the data structure of the second sample color feature C(2) is a grayscale value set, which includes the average grayscale value of multiple texture clustering clusters; at the same time, this embodiment does not specifically limit the calculation process of the average grayscale value WL(j) of each texture clustering cluster. Those skilled in the art can calculate the average grayscale value of the pixel points in each texture clustering cluster as the average grayscale value WL(j) of each texture clustering cluster.

[0091] Please continue reading Figure 1 As shown, the canine intelligent iris recognition system further includes a target image acquisition module, which is used to acquire a target iris image that meets a set standard;

[0092] The setting standard is: obtaining the shooting light intensity w, and collecting the target iris image according to the shooting light intensity w; the specific process is as follows: the target image acquisition module compares the shooting light intensity w with the preset shooting light intensity W, and collects the target iris image according to the comparison result: if W1≤w<W2, the target image acquisition module determines that the lighting environment is normal, and collects the target iris image; if w≥W2 or w<W1, the target image acquisition module determines that the lighting environment is abnormal, and does not collect the target iris image;

[0093] Wherein, W1 is the first preset light intensity, W2 is the second preset light intensity, W1<W2; the target image acquisition module collects images that meet the set standards to eliminate the influence of outdoor sclera reflection, so that the target iris image can be processed by the subsequent feature extraction process.

[0094] Specifically, this embodiment does not make any specific limitation on the value of the first preset light intensity W1 and the second preset light intensity W2. Those skilled in the art can set it freely, and only need to meet the value requirements of the first preset light intensity W1 and the second preset light intensity W2. For example, the first preset light intensity W1 can be set to 1000 lux, and the second preset light intensity W2 can be set to 10000 lux. The target iris image is an iris image of a canine animal in RGB format collected by the target. This embodiment does not make any specific limitation on the method for obtaining the shooting light intensity. Those skilled in the art can set it freely, and only need to meet the acquisition requirements of the shooting light intensity, such as obtaining it through a light intensity sensor. At the same time, the target iris image of this embodiment is collected by shooting with a high-definition camera.

[0095] Specifically, the shooting distance, shooting angle, pupil center position, image resolution and image format of the target iris image and the sample iris image obtained in this embodiment are the same. The implementation process is to aim the camera at the pupil center position through the above-mentioned sample pupil area division process, and the shooting distance is achieved through user control.

[0096] See also Figure 1 As shown, the intelligent iris recognition system for canines also includes a collection and screening module, which is used to analyze the saturation and contrast of the target iris image and adjust the collection process of the target iris image according to the saturation and contrast analysis results of the target iris image. The collection and screening module is connected to the target image collection module.

[0097] See also Figure 3 As shown, the acquisition and screening module is provided with a saturation analysis unit, and the saturation analysis unit is used to analyze the saturation of the target iris image and adjust the acquisition process of the target iris image according to the saturation analysis result;

[0098] The saturation analysis unit converts the target iris image into an HSV color image and calculates the target saturation index η, setting ;

[0099] Among them, max(R,G,B) represents the maximum value of the three color channels (R,G,B) in the HSV color image, min(R,G,B) represents the minimum value of the three color channels (R,G,B) in the HSV color image, and V is the brightness of the HSV color image;

[0100] The saturation analysis unit compares the target saturation index η with the saturation index threshold N, and adjusts the acquisition process of the target iris image according to the comparison result: if η≥N, the saturation analysis unit determines that the saturation is high, and adjusts the second preset light intensity to W2', setting W2'=W2×[(η-N) / N]; if η<N, the saturation analysis unit determines that the saturation is normal and does not make any adjustment; the saturation analysis unit calculates the image saturation and adjusts the set standard to incorporate the saturation into the acquisition standard, thereby improving the efficiency of the subsequent feature extraction process of the target iris image.

[0101] Specifically, this embodiment does not impose any specific limitation on the value of the saturation index threshold N, and those skilled in the art can freely set it as long as the value requirement of the saturation index threshold N is met. For example, the saturation index threshold N can be set to 0.7. At the same time, the process of "converting the target iris image into an HSV color image" in this embodiment is not described in detail. Those skilled in the art can implement the image format conversion through Python code, or through professional software (such as Photoshop).

[0102] Please continue reading Figure 3As shown, the acquisition and screening module also includes a contrast analysis unit, which is used to analyze the contrast of the target iris image and optimize the acquisition adjustment process of the target iris image according to the saturation analysis result; the contrast analysis unit optimizes the contrast of the target iris image by calculating the contrast of the target iris image;

[0103] The contrast analysis unit calculates the contrast C of the target iris image and sets , the contrast analysis unit optimizes the acquisition and adjustment process of the target iris image according to the contrast C, optimizes the saturation index threshold N to N', and sets N'=N×ln(1+C);

[0104] Wherein, T is the number of pixels of the target iris image, gmr(i) is the i-th pixel of the target iris image, and Pgmr is the mean grayscale value of the pixels of the target iris image; the contrast analysis unit calculates the contrast of the target iris image and optimizes the contrast of the target iris image according to the contrast calculation result of the target iris image, thereby improving the efficiency of subsequent feature extraction.

[0105] Please continue reading Figure 1 As shown, the canine intelligent iris recognition system is further provided with a target feature extraction module, which is used to extract the target canine iris features according to the target iris image;

[0106] The target feature extraction module analyzes the target edge feature, target texture feature and target color feature according to the target iris image.

[0107] Specifically, the process of "extracting target dog iris features according to target color" in this embodiment is the same as the implementation process of "extracting dog iris features according to sample iris images" mentioned above, and this embodiment will not be repeated here; that is, the analysis process of the target edge features is the same as the edge feature analysis process, the analysis results of the target texture features are the same as the analysis results of the texture features, and the analysis results of the target color features are the same as the analysis process of the color features.

[0108] See also Figure 1 As shown, the canine intelligent iris recognition system is further provided with a feature matching module, which is used to perform iris feature matching between the target canine iris feature extraction result and the canine iris feature extraction result, and the feature matching module is connected to the sample feature extraction module and the target feature extraction module.

[0109] See also Figure 4 As shown, the feature matching module includes an edge feature matching unit, which is used to match the target edge feature with the edge feature;

[0110] The edge feature matching unit calculates the edge similarity α according to the outer iris curve and the target outer iris curve, and sets α=Σ[(az-cz) 2 +(bz-dz) 2 ] 1 / 2 / K; where z is a mathematical subscript, set z=1,2...K, K is the number of pixels of the outer iris curve; az is the horizontal coordinate of the zth pixel in the target edge curve, cz is the horizontal coordinate of the zth pixel in the edge curve, bz is the vertical coordinate of the zth pixel in the target edge curve, and dz is the vertical coordinate of the zth pixel in the edge curve;

[0111] The edge feature matching unit compares the edge similarity α with the edge similarity threshold B, and analyzes the edge feature matching result according to the comparison result: if α<B, the edge feature matching unit determines that the edge feature matching is successful; if α≥B, the edge feature matching unit determines that the edge feature matching fails.

[0112] Specifically, this embodiment does not impose any specific limitation on the value of the edge similarity threshold B, and those skilled in the art can freely set it as long as the value requirement of the edge similarity threshold B is met. For example, the edge similarity threshold B can be set to 3.

[0113] Please continue reading Figure 4 As shown, the feature matching module is further provided with a texture feature matching unit, and the texture feature matching unit is used to perform texture feature matching on the target texture feature and the texture feature: when Ah≠ASh, the texture feature matching unit determines that the texture features do not match; when Ah=ASh=1, the texture feature matching unit calculates the first offset index γ1, and sets γ1=[q1(m)-q2(m)]·[qs1(m)-qs2(m)]; when Ah=ASh=2, the texture feature matching unit calculates the second offset index γ2, and sets γ2=Σ|wz(f)-wzs(f)| / F; when Ah=ASh=3, the texture feature matching unit calculates the third offset index γ3;

[0114] Among them, ASh represents the target recognition mode index, including AS1, AS2, AS3, and AS4, qs1(m) is the starting pixel coordinate of the mth radial line of the target, and qs2(m) is the ending pixel coordinate of the mth radial line of the target.

[0115] The texture feature matching unit compares each offset index γu with the preset offset index Y, sets u=1, 2, 3, and analyzes the texture matching result according to the comparison result: if γu≥Y, the texture feature matching unit determines that the texture matching fails; if γu<Y, the texture feature matching unit determines that the texture matching succeeds.

[0116] Specifically, the calculation process of the third offset index γ3 is the same as the calculation process of the edge similarity α. Those skilled in the art can substitute the two edge curves into the outer iris curve and the target outer iris curve respectively, which will not be elaborated in this embodiment. At the same time, in this embodiment, the value of the preset offset index Y is not specifically limited. Those skilled in the art can set it freely as long as the value requirement of the preset offset index Y is met. For example, the preset offset index Y can be set to 1. At the same time, the calculation process of "[q1(m)-q2(m)]·[qs1(m)-qs2(m)]" is specifically the dot product of the vector [q1(m)-q2(m)] and the vector [qs1(m)-qs2(m)].

[0117] Please continue reading Figure 4 As shown, the feature matching module is further provided with a color feature matching unit, and the color feature matching unit is used to perform color feature matching on the target color feature and the color feature: when C(1)=CS(1), the color feature matching unit determines that the first color feature is matched successfully. At this time, if C(2)=CS(2), the color feature matching unit determines that the color matching is successful; if C(2)≠CS(2), the color feature matching unit determines that the color matching fails; when C(1)≠CS(1), the color feature matching unit determines that the color feature matching fails;

[0118] Among them, CS(2) is the second color feature of the target, and CS(1) is the first color feature of the target.

[0119] Specifically, the equivalence judgment process of C(1) and CS(1) in this embodiment is a set equivalence judgment.

[0120] Please continue reading Figure 4 As shown, the feature matching module is further provided with an iris recognition unit, which analyzes the iris feature matching result according to the edge feature matching result, the texture feature matching result and the color feature matching result: if the edge feature matching is successful, the texture feature matching is successful and the color feature matching is successful, the iris recognition unit determines that the target iris image and the sample iris image are successfully matched; if the edge feature matching fails, the texture feature matching fails and the color feature matching fails, the iris recognition unit determines that the target iris image and the sample iris image are unsuccessfully matched; the iris recognition unit analyzes each feature matching result to judge the current recognition result, thereby improving recognition efficiency and accuracy.

[0121] Please continue to participate Figure 1 As shown, the canine intelligent iris recognition system is further provided with an iterative matching module, which is used to iterate the iris feature matching process, and the iterative matching module is connected to the feature matching module;

[0122] The iterative matching module iterates the iris feature matching process according to the canine iris features stored in the server: when there are still unmatched canine iris features in the server, if the iris feature matching result of the current canine iris feature is a failure, the iterative matching module uses the next canine iris feature as the canine iris feature extraction result of the next iris matching process; if the iris feature matching result of the current canine iris feature is a success, the iterative matching module stops the iterative process and sends the iris feature matching result to the user; when there are no unmatched canine iris features in the server, if the iris feature matching result of the current canine iris feature is a failure, the iterative matching module sends "matching failure" to the user; if the iris feature matching result of the current canine iris feature is a success, the iterative matching module stops the iterative process and sends the iris feature matching result to the user; the iterative matching module improves the efficiency of matching judgment by matching the canine iris features stored in the server one by one and setting a stop event.

[0123] Specifically, the current canine iris feature is the canine iris feature that is currently undergoing iris feature matching; the unmatched canine iris feature is the canine iris feature that has not yet undergone iris feature matching; it can be understood that the server of this embodiment stores canine iris features of multiple canine animals, and the number of such features is the same as the number of sample iris images entered by the user.

[0124] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. An intelligent iris recognition system for canines, characterized in that: include: A sample image acquisition module, used for acquiring a sample iris image; A sample feature extraction module, used to divide the sample iris image into regions, and extract edge features, texture features and color features according to the region division results, and use the extraction results as canine iris features. The sample feature extraction module is also used to store the canine iris features in a server; A target image acquisition module is used to acquire a target iris image that meets the set standards; An acquisition and screening module is used to analyze the saturation and contrast of the target iris image, and adjust the acquisition process of the target iris image according to the analysis results of the saturation and contrast of the target iris image; A target feature extraction module is used to extract the iris features of the target dog according to the target iris image; A feature matching module, used for performing iris feature matching between the target dog iris feature extraction result and the dog iris feature extraction result; An iterative matching module, used to iterate the iris feature matching process and send the iterative result of the iris feature matching to the user; The sample feature extraction module is further provided with an edge feature extraction unit, and the edge feature extraction unit analyzes the edge feature according to the region division result of the sample iris image; The edge feature extraction unit uses the intersection pixel points of the sample iris area and the sample pupil area as the inner iris edge curve pixel points, and uses the intersection pixel points of the sample sclera area and the sample iris area as the outer iris edge pixel points; The edge feature extraction unit connects the inner iris pixel points into an inner iris curve and connects the outer iris pixel points into an outer iris curve; The edge feature extraction unit judges the sample iris image according to the analysis result of the inner iris curve: if the inner iris curve is a circular curve, the edge feature extraction unit judges that the sample iris image is qualified; if the inner iris curve is a non-circular curve, the edge feature extraction unit judges that the sample iris image is unqualified; The edge feature extraction unit stores the outer iris curve as an edge feature.

2. The canine intelligent iris recognition system according to claim 1, characterized in that: The sample feature extraction module is provided with a region division unit, and the region division unit is used to divide the sample iris image into regions; The region division unit clusters the pixels of the iris image according to the grayscale values ​​g(x,y) of the pixels of the sample iris image to obtain clusters; Where g(x,y) is the gray value of the pixel whose horizontal coordinate is x and whose vertical coordinate is y; The region division unit divides the pixels included in each cluster into continuous regions according to the coordinates of the pixels included in each cluster: the region division unit combines the pixels included in the same continuous cluster into divided regions; The region division unit classifies each divided region according to the average gray value Gray(i) of each divided region: if Gray(i)∈{[0,5],[0,5],[0,5]}, the region division unit determines that the divided region is the sample pupil region; if Gray(i)∈{[200,255],[240,255],[240,255]}, the region division unit determines that the divided region is the sample sclera region; if g(x,y)∈{[0,5],[0,5],[0,5]} and g(x,y) {[200,255],[240,255],[240,255]}, the region division unit takes the pixel point as the sample iris pixel point, and combines the sample iris pixel points into a sample iris region; the region division unit determines the region composed of the remaining pixel points as the remaining region.

3. The canine intelligent iris recognition system according to claim 2, characterized in that: The sample feature extraction module is further provided with a texture feature extraction unit, which extracts texture features of the sample iris according to the judgment result and the area division result of the sample iris image; The texture feature extraction unit performs grayscale processing on the sample iris area: the texture feature analysis unit sets the grayscale value of the pixel points in the sample iris area to G(x,y), and sets G(x,y)=|g(x,y)|; The texture feature extraction unit clusters the pixel points of the sample iris area according to the grayscale values ​​of the pixel points to obtain various texture clusters; The texture feature analysis unit connects edge curves s(j) of each texture cluster according to the pixel points in each texture cluster; The texture feature extraction unit judges the sample iris pattern feature according to the edge curve s(j) of each texture cluster, and sets the recognition mode index according to the judgment result of the sample iris shape feature: if the sample iris pattern feature is radial, the texture feature extraction unit sets the recognition mode index to A1, sets A1=1, and uses the radial starting point pixel coordinate q1(m) and the radial end point pixel coordinate q2(m) as the sample texture data; if the sample iris pattern feature is mesh, the texture feature extraction unit sets the recognition mode index to A2, sets A2=2, and uses the mesh node wz(f) as the sample texture data; if the sample iris pattern feature is other, the texture feature extraction unit sets the recognition mode index to A3, sets A3=3, and uses the edge curve s(j) of each texture cluster as the sample texture data; Wherein, q1(m) is the starting pixel coordinate of the m-th radial line, q2(m) is the ending pixel coordinate of the m-th radial line, wz(f) is the pixel coordinate of the f-th mesh node, and m and f are both digital subscripts; The texture feature extraction unit uses the recognition pattern index Ah and the sample texture data as the texture feature extraction result, and sets h=1, 2, 3.

4. The canine intelligent iris recognition system according to claim 3, characterized in that: The sample feature extraction module is further provided with a color feature extraction unit, and the color feature extraction unit is used to extract the sample color feature according to the area division result of the sample iris image; The color feature extraction unit uses the average gray value Gray(i) of the sample pupil area as the first sample color feature C(1); The color feature extraction unit uses the average grayscale value WL(j) of each texture cluster of the sample iris region as the second sample color feature C(2); The color feature extraction unit uses the first color feature, the second color feature, and the texture feature extraction results as feature extraction results of the sample iris image.

5. The canine intelligent iris recognition system according to claim 4, characterized in that: The target image acquisition module compares the shooting light intensity w with the preset shooting light intensity W, and acquires the target iris image according to the comparison result: if W1≤w<W2, the target image acquisition module determines that the lighting environment is normal and acquires the target iris image; if w≥W2 or w<W1, the target image acquisition module determines that the lighting environment is abnormal and does not acquire the target iris image; Wherein, W1 is the first preset light intensity, W2 is the second preset light intensity, and W1<W2.

6. The canine intelligent iris recognition system according to claim 5, characterized in that: The acquisition and screening module is provided with a saturation analysis unit, and the saturation analysis unit is used to analyze the saturation of the target iris image and adjust the acquisition process of the target iris image according to the saturation analysis result; The saturation analysis unit converts the target iris image into an HSV color image and calculates the target saturation index η, setting ; Among them, max(R,G,B) represents the maximum value of the three color channels (R,G,B) in the HSV color image, min(R,G,B) represents the minimum value of the three color channels (R,G,B) in the HSV color image, and V is the brightness of the HSV color image; The saturation analysis unit compares the target saturation index η with the saturation index threshold N, and adjusts the acquisition process of the target iris image according to the comparison result: if η≥N, the saturation analysis unit determines that the saturation is high, and adjusts the second preset light intensity to W2'; if η<N, the saturation analysis unit determines that the saturation is normal and does not make any adjustment; The acquisition and screening module is also provided with a contrast analysis unit, which is used to analyze the contrast of the target iris image and optimize the acquisition and adjustment process of the target iris image according to the saturation analysis result; The contrast analysis unit calculates the contrast C of the target iris image and sets , the contrast analysis unit optimizes the acquisition and adjustment process of the target iris image according to the contrast C, and optimizes the saturation index threshold N to N'; Wherein, T is the number of pixels of the target iris image, gmr(i) is the i-th pixel of the target iris image, and Pgmr is the mean grayscale value of the pixels of the target iris image.

7. The canine intelligent iris recognition system according to claim 6, characterized in that: The feature matching module is provided with an edge feature matching unit, which is used to match the target edge feature with the edge feature; The edge feature matching unit calculates the edge similarity α according to the outer iris curve and the target outer iris curve, and sets α=Σ[(az-cz) 2 +(bz-dz) 2 ] 1 / 2 / K; where z is a mathematical subscript, set z=1,2...K, K is the number of pixels of the outer iris curve; az is the horizontal coordinate of the zth pixel in the target edge curve, cz is the horizontal coordinate of the zth pixel in the edge curve, bz is the vertical coordinate of the zth pixel in the target edge curve, and dz is the vertical coordinate of the zth pixel in the edge curve; The edge feature matching unit compares the edge similarity α with the edge similarity threshold B, and analyzes the edge feature matching result according to the comparison result: if α<B, the edge feature matching unit determines that the edge feature matching is successful; if α≥B, the edge feature matching unit determines that the edge feature matching fails.

8. The canine intelligent iris recognition system according to claim 7, characterized in that: The texture feature matching unit is used to perform texture feature matching on the target texture feature and the texture feature: when Ah≠ASh, the texture feature matching unit determines that the texture features do not match; when Ah=ASh=1, the texture feature matching unit calculates the first offset index γ1, and sets γ1=[q1(m)-q2(m)]·[qs1(m)-qs2(m)]; when Ah=ASh=2, the texture feature matching unit calculates the second offset index γ2, and sets γ2=Σ|wz(f)-wzs(f)| / F; when Ah=ASh=3, the texture feature matching unit calculates the third offset index γ3; Among them, ASh represents the index of each target recognition mode, including AS1, AS2, AS3, and AS4; The texture feature matching unit compares each offset index γu with a preset offset index Y, sets u=1, 2, 3, and analyzes the texture matching result according to the comparison result: if γu≥Y, the texture feature matching unit determines that the texture matching fails; if γu<Y, the texture feature matching unit determines that the texture matching succeeds; The feature matching module is further provided with a color feature matching unit, which is used to perform color feature matching on the target color feature and the color feature: when C(1)=CS(1), the color feature matching unit determines that the first color feature is matched successfully. At this time, if C(2)=CS(2), the color feature matching unit determines that the color matching is successful; if C(2)≠CS(2), the color feature matching unit determines that the color matching fails; when C(1)≠CS(1), the color feature matching unit determines that the color feature matching fails; Among them, CS(2) is the second color feature of the target, and CS(1) is the first color feature of the target.

9. The canine intelligent iris recognition system according to claim 8, characterized in that: The iterative matching module is used to iterate the iris feature matching process; The iterative matching module iterates the iris feature matching process according to the canine iris features stored in the server: when there are still unmatched canine iris features in the server, if the iris feature matching result of the current canine iris feature is a failure, the iterative matching module uses the next canine iris feature as the canine iris feature extraction result of the next iris matching process; if the iris feature matching result of the current canine iris feature is a success, the iterative matching module stops the iterative process and sends the iris feature matching result to the user; when there are no unmatched canine iris features in the server, if the iris feature matching result of the current canine iris feature is a failure, the iterative matching module sends "matching failure" to the user; if the iris feature matching result of the current canine iris feature is a success, the iterative matching module stops the iterative process and sends the iris feature matching result to the user.

Images