A palmprint recognition unlocking system and method based on image recognition

Abstract

The application relates to a palmprint recognition unlocking system and method based on image recognition, in particular to the technical field of image recognition, which comprises an input module used for inputting a target user's palmprint, taking the obtained target user's palmprint as a palmprint training sample, the input module is also used for classifying palmprint test samples, and the input module is also used for training the palmprint training sample according to the palmprint test sample; a collection module used for collecting a user's palmprint in real time; a storage module used for storing and classifying the input palmprint training sample and the collected user's palmprint; a matching module used for matching the collected user's palmprint with the stored target user's palmprint, and calculating the similarity between the user's palmprint and the target user's palmprint; a control module used for controlling the opening of a vehicle door when the target user's palmprint is matched successfully; and an alarm module used for alarming when the target user's palmprint is not matched successfully. The application effectively improves the safety and efficiency of vehicle door unlocking.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and in particular to a palmprint recognition unlocking system and method based on image recognition. Background Technology

[0002] With continuous technological advancements, people are increasingly focusing on intelligence, convenience, comfort, and security when using vehicles, leading to research into various car door unlocking methods. Car fingerprint locks, also known as car fingerprint anti-theft devices, are devices that utilize fingerprint recognition technology for car anti-theft. They can control the car's electrical circuits, fuel system, engine starter motor, etc., via fingerprints, and are characterized by their anti-jamming and anti-decoding features. Car fingerprint locks have the advantages of high recognition rate and low false recognition rate, and are generally applied to various aspects of automobiles.

[0003] Compared to fingerprint recognition, palmprint recognition is less affected by physiological and environmental factors, such as skin moisture, cleanliness, peeling, and scars. Palmprints are also more difficult to forge, making them more secure. In addition, palmprint recognition is faster and has a higher recognition rate. Summary of the Invention

[0004] Therefore, the present invention provides a palmprint recognition unlocking system and method based on image recognition, in order to overcome the problems of low security and low recognition efficiency of fingerprint recognition compared with palmprint recognition in the prior art.

[0005] To achieve the above objectives, in one aspect, the present invention provides a palmprint recognition unlocking system based on image recognition, comprising:

[0006] The input module is used to input the palmprint of the target user, using the acquired palmprint as a palmprint training sample. The input module also preprocesses the palmprint training sample and extracts features from it to obtain the target palmprint features. Furthermore, the input module classifies palmprint test samples. During classification, the input module preprocesses and extracts features from the palmprint test samples. After feature extraction, the input module feeds the palmprint test samples into a classifier for classification. After classification, the input module trains the palmprint training sample based on the palmprint test samples until the palmprint similarity between the palmprint test samples and the palmprint training samples meets the requirements.

[0007] The data acquisition module is used to collect users' palm prints in real time.

[0008] The storage module is used to store and classify the input palmprint training samples and the collected user palmprints;

[0009] The matching module is used to match the collected user palm prints with the stored target user palm prints and calculate the similarity between the user palm prints and the target user palm prints.

[0010] The control module is used to open the car door when the target user's palm print is successfully matched;

[0011] The alarm module is used to issue an alarm when the target user's palm print fails to match.

[0012] Furthermore, when the input module preprocesses the palmprint training samples, the preprocessing process includes grayscale processing, geometric transformation processing, and image enhancement processing.

[0013] Furthermore, during the grayscale processing, the input module uses the component method, maximum value method, average value method, and weighted average method to perform grayscale processing on the palmprint training samples.

[0014] Furthermore, when performing geometric transformation processing on the palmprint training samples, the input module processes the palmprint training samples through geometric transformations such as translation, transpose, mirroring, rotation, and scaling to correct systematic errors and random errors in instrument position during the input of palmprint training samples; when performing image enhancement processing on the palmprint training samples, the input module employs spatial domain methods and frequency domain methods.

[0015] Furthermore, when the input module extracts features from the palmprint training samples, it uses an encoding-based method to extract features, treating the palmprint training samples as texture images and encoding the texture images for feature extraction.

[0016] Furthermore, when encoding the texture image, the input module uses a set of 2D Gabor wavelet filters with uniform scale but different directions to convolve the image to obtain the response value corresponding to each pixel of the texture image, and approximates the palm line direction of the point by using the filter direction corresponding to the maximum response value.

[0017] Furthermore, when classifying palmprint test samples, the classifier uses a probability density-based classification algorithm that leverages Bayesian theory and employs a latent class-conditional probability density function to classify the palmprint test samples.

[0018] Furthermore, when calculating the palmprint matching similarity between the palmprint test sample and the palmprint training sample, the input module uses angular distance to determine the similarity between the two palmprint images. The requirement is met when the similarity is greater than the threshold.

[0019] Furthermore, when performing matching, the matching module compares the similarity between the user's palm print and the target user's palm print with a threshold. When the similarity between the user's palm print and the target user's palm print is greater than the threshold, the matching module determines that the target user's palm print match is successful. When the similarity between the user's palm print and the target user's palm print is less than the threshold, the matching module determines that the target user's palm print match fails.

[0020] On the other hand, the present invention also provides a palmprint recognition unlocking method based on image recognition, comprising:

[0021] Step S1: Input the target user's palm print, obtain palm print training samples and palm print test samples respectively, preprocess and extract features from the palm print training samples to obtain the target palm print features, and preprocess, extract features and classify the palm print test samples.

[0022] Step S2: Train the palmprint training sample based on the palmprint test sample until the palmprint matching similarity between the palmprint test sample and the palmprint training sample meets the requirements.

[0023] Step S3: Collect user palm prints in real time and match the collected user palm prints with the stored target user palm prints;

[0024] Step S4: Control the car door to open when the target user's palm print is successfully matched, and issue an alarm when the target user's palm print is not matched.

[0025] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention trains the input palmprint training samples to ensure that the matching degree between the trained palmprint training samples and the user's palmprint meets the requirements, thereby facilitating palmprint recognition. By storing the trained palmprint training samples as the target user's palmprint and matching the user's real-time collected palmprint with them, the door is unlocked when the matching result meets the requirements, and an alarm is triggered otherwise. Mismatched palmprints can be recorded and displayed on the infotainment system, and an alarm can be triggered to alert the owner that someone is trying to illegally open the door, thereby realizing the vehicle unlocking function and improving the convenience, security, and efficiency of vehicle door unlocking. This invention unlocks the vehicle by palmprint recognition, mainly by storing the trained palmprint model in the system and comparing the scanned palmprint with the palmprint stored in the system database, thereby quickly unlocking the door and providing convenience for the user. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the palmprint recognition unlocking system based on image recognition in this embodiment;

[0027] Figure 2 This is a flowchart illustrating the palmprint recognition unlocking method based on image recognition in this embodiment. Detailed Implementation

[0028] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0029] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0030] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0031] Please see Figure 1 As shown, this is a palmprint recognition unlocking system based on image recognition in this embodiment. The system includes:

[0032] The input module is used to input the palmprint of the target user, using the acquired palmprint as a training sample. The input module also preprocesses the training sample and extracts features from it to obtain the target palmprint features. The input module further classifies palmprint test samples. During classification, the input module preprocesses and extracts features from the test samples. After feature extraction, the input module feeds the test samples into a classifier for classification. After classification, the input module... The input module is also used to train the palmprint training sample based on the palmprint test sample until the palmprint matching similarity between the palmprint test sample and the palmprint training sample meets the requirements. In this embodiment, the palmprint training sample is the palmprint entered for the first time when the target user enters the palmprint, and the palmprint test sample is the palmprint entered after the first time when the target user enters the palmprint. For example, if a total of 5 palmprints are entered when entering the palmprint, the first one is the palmprint training sample and the last 4 are the palmprint test samples. This embodiment does not make a specific limit on the number of palmprints entered when entering the palmprint, and those skilled in the art can set it freely.

[0033] The acquisition module is used to acquire the user's palm print in real time; in this embodiment, the palm print is acquired in real time by installing a palm print recognition device on the car door, and the acquisition module is connected to the input module.

[0034] The storage module is used to store and classify the input palmprint training samples and the collected user palmprints, including target user palmprints and ordinary user palmprints. The storage module is connected to the collection module. In this embodiment, the target user palmprint is the input palmprint training sample of the target user, and the ordinary user palmprint is the collected palmprint of the non-target user.

[0035] The matching module is used to match the collected user palmprint with the stored target user palmprint and calculate the similarity between the two. When the similarity between the user palmprint and the target user palmprint is greater than a threshold, the matching module determines that the target user palmprint is successfully matched; when the similarity is less than the threshold, the matching module determines that the target user palmprint is not matched. The matching module is connected to the storage module. This embodiment does not specifically limit the similarity calculation method. Those skilled in the art can freely set it, such as using the EDCC algorithm for similarity calculation, as long as the similarity calculation requirements are met.

[0036] A control module is used to control the car door to open when the target user's palm print is successfully matched; the control module is connected to the matching module.

[0037] The alarm module is used to issue an alarm when the palmprint matching of the target user fails. The alarm module is connected to the matching module. This embodiment does not specifically limit the alarm method. Those skilled in the art can freely set it, such as setting it to a sound alarm.

[0038] Specifically, this invention trains the input palmprint training samples to ensure that the trained palmprint training samples match the user's palmprint to a certain degree, thus facilitating palmprint recognition. The trained palmprint training samples are stored as the target user's palmprint, and real-time collected user palmprints are matched with them. When the matching result meets the requirements, the car door is unlocked; otherwise, an alarm is triggered. Mismatched palmprints can be recorded and displayed on the infotainment system, and an alarm can be triggered to alert the owner that someone is attempting to illegally open the door, thereby achieving the vehicle unlocking function and improving the convenience and security of unlocking. This invention unlocks vehicles through palmprint recognition, mainly by storing a trained palmprint model in the system. The scanned palmprint is compared with the palmprints stored in the system database to quickly unlock the car door, providing convenience for the user.

[0039] Specifically, in this embodiment, when the input module preprocesses the palmprint training samples, the preprocessing process includes grayscale processing, geometric transformation processing, and image enhancement processing. When performing grayscale processing, any one of the following methods can be used to perform grayscale processing on the palmprint training samples: component method, maximum value method, average value method, and weighted average method.

[0040] Specifically, in this embodiment, when using the component method for grayscale processing, the brightness of three components in the palmprint training sample is used as the grayscale values ​​of the three grayscale images. One grayscale image can be selected according to application needs, and the following settings are made:

[0041] f1(i,j)=R(i,j), f2(i,j)=G(i,j), f3(i,j)=B(i,j)

[0042] Where fk(i,j) (k=1,2,3) is the gray value of the converted grayscale image at (i,j).

[0043] Specifically, in this embodiment, when using the maximum value method for grayscale processing, the maximum value of the three brightness components in the palmprint training sample is taken as the grayscale value of the grayscale image, and the following settings are made:

[0044] f(i,j)=max(R(i,j),G(i,j),B(i,j)).

[0045] Specifically, in this embodiment, when using the averaging method for grayscale processing, the three components of brightness in the palmprint training sample are averaged to obtain a grayscale value, and the following settings are made:

[0046] f(i,j)=(R(i,j)+G(i,j)+B(i,j)) / 3.

[0047] Specifically, in this embodiment, when using the weighted average method for grayscale processing, the three components are weighted and averaged with different weights based on their importance and other indicators. Since the human eye is most sensitive to green and least sensitive to blue, a more reasonable grayscale image can be obtained by weighting and averaging the RGB components according to the following formula:

[0048] f(i,j)=0.30R(i,j)+0.59G(i,j)+0.11B(i,j).

[0049] Specifically, image geometric transformation, also known as image spatial transformation, is used in this embodiment. When processing palmprint training samples, the input module performs geometric transformations such as translation, transpose, mirroring, rotation, and scaling to correct systematic errors and random errors in instrument positioning during palmprint training sample input. It is worth noting that grayscale interpolation algorithms are also required during geometric transformation processing. This is because, according to this transformation relationship, the pixels of the output image may be mapped to non-integer coordinates of the input image. Commonly used methods include nearest neighbor interpolation, bilinear interpolation, and bicubic interpolation.

[0050] Specifically, in this embodiment, the input module performs geometric transformation processing on the palmprint training samples, followed by image enhancement processing to enhance the useful information in the image. Image enhancement is a distortion process aimed at improving the visual effect of the image. Depending on the application context of the given image, it purposefully emphasizes the overall or local characteristics of the image, making an originally unclear image clearer or emphasizing certain features of interest, amplifying the differences between features of different objects in the image, and suppressing features of little interest. This improves image quality, enriches information, enhances image interpretation and recognition, and meets the needs of certain special analyses. In this embodiment, when performing image enhancement processing on the palmprint training samples, spatial domain methods or frequency domain methods can be used. Spatial domain-based algorithms directly operate on the image grayscale levels, while frequency domain-based algorithms correct the transformation coefficients of the image within a certain transformation domain, which is an indirect enhancement algorithm.

[0051] Specifically, spatial domain methods are direct image enhancement algorithms, divided into point operation algorithms and neighborhood enhancement algorithms. Point operation algorithms include gray-level correction, gray-level transformation, and histogram correction, aiming to either make the image more uniform or expand the dynamic range and contrast. Neighborhood enhancement algorithms include image smoothing and sharpening. Smoothing is generally used to eliminate image noise, but it can easily cause edge blurring. Common smoothing algorithms include mean filtering, median filtering, and spatial domain filtering. Sharpening aims to highlight the edge contours of objects for easier target recognition. Common sharpening algorithms include gradient operator methods, second derivative operator methods, high-pass filtering, and mask matching methods. Spatial domain methods operate on the pixels in the image, described by the following formula:

[0052] g(x,y)=f(x,y)*h(x,y)

[0053] Where f(x,y) is the original image, h(x,y) is the spatial transformation function, and g(x,y) represents the processed image.

[0054] Specifically, frequency domain methods are indirect image enhancement algorithms. Commonly used frequency domain enhancement methods include low-pass filters and high-pass filters. Low-pass filters include ideal low-pass filters, Butterworth low-pass filters, Gaussian low-pass filters, and exponential filters, while high-pass filters include ideal high-pass filters, Butterworth high-pass filters, Gaussian high-pass filters, and exponential filters. Frequency domain methods treat the image as a two-dimensional signal and perform signal enhancement based on two-dimensional Fourier transform. When using low-pass filtering, noise in the image can be removed; when using high-pass filtering, high-frequency signals such as edges can be enhanced, making blurry images clearer.

[0055] Specifically, in this embodiment, the input module uses an encoding-based method to extract features from palmprint training samples, treating the palmprint training samples as texture images and encoding the texture images for feature extraction.

[0056] Specifically, in this embodiment, the input module encodes the texture image by convolving a set of 2D Gabor wavelet filters with uniform scale but different directions to obtain the response value corresponding to each pixel in the texture image. The filter direction corresponding to the maximum response value approximates the palmprint line direction at that point. The method of extracting the main direction of the palmprint based on the filter response is based on the fundamental assumption that pixels in the palmprint image belong to a line. When the direction of the filter is the same as the main direction of the palmprint, the filter response will reach its maximum value. It is worth noting that the closer the direction of the 2D Gabor wavelet filter is to the main direction of the palmprint, the larger the obtained filter response. Therefore, the extracted adjacent directions can be combined with the main direction to more accurately represent the palmprint image. It is understood that the filter response is proportional to the degree of overlap between the main line and the filter. Since the filter directions used in practice are limited, there may not be a filter with the same orientation as the main direction of the palmprint image. In this case, the extracted direction cannot accurately represent the main direction feature of the palmprint, and feature extraction needs to be performed again.

[0057] Specifically, in this embodiment, the classifier learns classification rules based on training data with known categories and classifies unknown data. In this embodiment, when classifying palmprint test samples, the classifier uses a probability density-based classification algorithm with the help of Bayesian theory and employs a latent class-conditional probability density function to classify palmprint test samples. At the same time, this embodiment does not impose specific limitations on the probability density classification algorithm, which can be freely set by those skilled in the art. For example, Bayesian estimation or maximum likelihood estimation can be used. These algorithms are parametric estimations, which require pre-assuming a distribution model of the categories and then using training data to adjust the various parameters in the probability density. Parametric estimation methods such as Parzen window and Kn nearest neighbor can also be used, which can directly estimate the probability density from the training samples.

[0058] Specifically, in this embodiment, the input module uses angular distance to determine the similarity between two palm print images when calculating the palm print matching similarity between the palm print test sample and the palm print training sample. The requirement is met when the similarity is greater than a threshold. In this embodiment, the similarity threshold is 90%.

[0059] Please see Figure 2 As shown, this is a palmprint recognition unlocking method based on image recognition in this embodiment. The method includes:

[0060] Step S1: Input the target user's palm print, obtain palm print training samples and palm print test samples respectively, preprocess and extract features from the palm print training samples to obtain the target palm print features, and preprocess, extract features and classify the palm print test samples.

[0061] Step S2: Train the palmprint training sample based on the palmprint test sample until the palmprint matching similarity between the palmprint test sample and the palmprint training sample meets the requirements.

[0062] Step S3: Collect user palm prints in real time and match the collected user palm prints with the stored target user palm prints;

[0063] Step S4: Control the car door to open when the target user's palm print is successfully matched, and issue an alarm when the target user's palm print is not matched.

[0064] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles 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 all fall within the scope of protection of the present invention.

Claims

1. An image recognition based palmprint recognition unlocking system, characterized in that, The application relates to a palmprint recognition system for vehicle door, which comprises the following modules: An input module is used to input target user palmprints, and the obtained target user palmprints are used as palmprint training samples; the input module is also used to pre-process the palmprint training samples and extract features of the palmprint training samples after the pre-processing. The input module is also used to classify palmprint test samples; when the classification is performed, the input module is also used to pre-process and extract features of the palmprint test samples; after the feature extraction of the palmprint test samples, the input module is also used to send the palmprint test samples to a classifier for classification; after the classification is completed, the input module is also used to train the palmprint training samples according to the palmprint test samples until the palmprint matching similarity of the palmprint test samples and the palmprint training samples meets the requirements. A collection module is used to collect user palmprints in real time. A storage module is used to store and classify the input palmprint training samples and the collected user palmprints. A matching module is used to match the collected user palmprints with the stored target user palmprints, and calculate the similarity of the user palmprints and the target user palmprints. A control module is used to control the opening of a vehicle door when the target user palmprints are matched successfully. An alarm module is used to alarm when the target user palmprints are not matched successfully. When the texture image is encoded, the input module adopts a group of 2DGabor wavelet filters with unified scales and different directions to perform convolution on the image, obtains the response value corresponding to each pixel point of the texture image, and approximately represents the palmprint line direction of the point by the filter direction corresponding to the maximum response value. When the palmprint test samples are classified, the classifier adopts a classification algorithm based on probability density, uses the Bayesian theory system, and adopts a latent class conditional probability density function to classify the palmprint test samples. When the palmprint matching similarity of the palmprint test samples and the palmprint training samples is calculated, the input module adopts an angle distance to determine the similarity between two palmprint images, and the requirements are met when the similarity is greater than a threshold value.

2. The image recognition based palmprint identification unlocking system according to claim 1, wherein, When the palmprint training samples are pre-processed, the pre-processing process comprises grayscale processing, geometric transformation processing and image enhancement processing.

3. The image recognition based palmprint identification unlocking system according to claim 2, wherein, When the grayscale processing is performed, the input module adopts a component method or a maximum value method or an average value method or a weighted average method to perform the grayscale processing on the palmprint training samples.

4. The image recognition based palmprint identification unlocking system according to claim 3, wherein, When the geometric transformation processing is performed on the palmprint training samples, the palmprint training samples are processed through translation, transposition, mirroring, rotation and scaling geometric transformation to correct system errors and random errors of instrument positions when the palmprint training samples are input; when the image enhancement processing is performed on the palmprint training samples, a spatial domain method and a frequency domain method are adopted.

5. The image recognition based palmprint identification unlocking system according to claim 4, wherein, When the feature extraction is performed on the palmprint training samples, an encoding-based method is adopted to extract features; the palmprint training samples are used as texture images, and the texture images are encoded to extract features.

6. The image recognition based palmprint identification unlocking system according to claim 1, wherein, The matching module compares the similarity between the user palmprint and the target user palmprint with a threshold value when performing matching, determines that the target user palmprint is matched successfully when the similarity between the user palmprint and the target user palmprint is greater than the threshold value, and determines that the target user palmprint is matched unsuccessfully when the similarity between the user palmprint and the target user palmprint is less than the threshold value.

7. A method applied to the palmprint recognition unlocking system based on image recognition according to any one of claims 1-6, characterized in that, Comprise: Step S1: inputting a target user palmprint, obtaining palmprint training samples and palmprint test samples respectively, pre-processing and feature extraction on the palmprint training samples to obtain target palmprint features, pre-processing, feature extraction and classification on the palmprint test samples; Step S2: training the palmprint training samples according to the palmprint test samples until the palmprint matching similarity between the palmprint test samples and the palmprint training samples meets the requirements; Step S3: real-time collection of a user palmprint and matching of the collected user palmprint with the stored target user palmprint; Step S4: controlling the opening of the vehicle door when the target user palmprint is matched successfully, and performing an alarm when the target user palmprint is matched unsuccessfully.

Images