Sensor-induced image recognition method, system, device and storage medium

Abstract

The present invention discloses a sensor-induced image recognition method, system, device, and storage medium. The method includes: obtaining first array data of a sensor-induced image, adding Gaussian noise to the first array data to obtain second array data; performing functional relationship conversion on the second array data to obtain a pixel capacitance array, and inputting the pixel capacitance array into a pre-trained convolutional neural network; performing convolution calculation on the pixel capacitance array through the convolution layer of the convolutional neural network to obtain a sensor charge array; performing data scaling on the sensor charge array to obtain third array data, extracting data feature information of the third array data through the activation function of the convolutional neural network; inputting the data feature information into the fully connected layer of the convolutional neural network to calculate an output vector, and determining the recognition result of the sensor-induced image based on the output vector. The present invention improves the recognition efficiency of sensor-induced images and can be widely applied in the field of artificial intelligence technology.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a sensor-induced image recognition method, system, device and storage medium. Background Art

[0002] Currently, common applications in edge capacitance pixel array sensors, such as fingerprint sensors, are to use the parasitic capacitance generated by capacitance pixels under external perception to change the capacitance value. The changed capacitance value is then measured by a measurement circuit and converted into a digital quantity. The processor then performs a classification algorithm on the digital data representing the capacitance value information obtained for each capacitance pixel to obtain the recognition result. This process does not fully utilize the electrical characteristics of the capacitor and the physical laws of the circuit structure, requires a large storage space and high computing power, and consumes a lot of computing time during the recognition process, affecting the recognition efficiency of the sensor-sensed image. Summary of the Invention

[0003] The purpose of the present invention is to solve one of the technical problems existing in the prior art to at least a certain extent.

[0004] To this end, one purpose of an embodiment of the present invention is to provide a sensor-induced image recognition method, which utilizes the electrical characteristics of capacitors and uses analog capacitance values ​​to participate in calculations, which can effectively simplify circuit resources, speed up the calculation process in the neural network, and realize rapid recognition of sensor-induced images. It is also beneficial to speed up the calculation speed, reduce data bandwidth, and simplify hardware resources during hardware mapping, thereby reducing the requirements for system storage space and computing power, and improving the recognition efficiency of sensor-induced images.

[0005] Another object of an embodiment of the present invention is to provide a sensor-induced image recognition system.

[0006] In order to achieve the above technical objectives, the technical solutions adopted by the embodiments of the present invention include:

[0007] In a first aspect, an embodiment of the present invention provides a sensor-induced image recognition method, comprising the following steps:

[0008] Acquire first array data of a sensor-sensed image, and add Gaussian noise to the first array data to obtain second array data;

[0009] Performing a functional relationship conversion on the second array data to obtain a pixel capacitance array, and then inputting the pixel capacitance array into a pre-trained convolutional neural network;

[0010] Performing convolution calculation on the pixel capacitance array through the convolution layer of the convolutional neural network to obtain a sensor charge array;

[0011] Performing data scaling on the sensor charge array to obtain third array data, and extracting data feature information of the third array data through the activation function of the convolutional neural network;

[0012] The data feature information is input into the fully connected layer of the convolutional neural network to calculate an output vector, and then the recognition result of the sensor-sensed image is determined according to the output vector.

[0013] Furthermore, in one embodiment of the present invention, the second array data is calculated by the following formula:

[0014] P(ij)=[1+σ(Noise)×randn(0,1)]Q(ij)

[0015] Wherein, P(ij) represents the data in the i-th row and j-th column of the second array data, Q(ij) represents the data in the i-th row and j-th column of the first array data, σ(Noise) represents the preset noise coefficient, and randn(0,1) represents a random number generated by a Gaussian function with a center of 0 and a variance of 1.

[0016] Furthermore, in one embodiment of the present invention, the pixel capacitance array is calculated by the following formula:

[0017] C(ij)=[P(ij)×C0] / [P(ij)+C0]

[0018] Wherein, C(ij) represents the data of the i-th row and j-th column in the pixel capacitance array, P(ij) represents the data of the i-th row and j-th column in the second array data, and C0 represents the original capacitance of the sensor.

[0019] Furthermore, in one embodiment of the present invention, the step of performing convolution calculation on the pixel capacitance array through the convolution layer of the convolutional neural network to obtain the sensor charge array is specifically:

[0020] Performing convolution calculation on the pixel capacitance array according to a preset first weight parameter to obtain the sensor charge array;

[0021] The first weight parameter is used to represent the voltage value applied to the sensor pixel capacitance.

[0022] Furthermore, in one embodiment of the present invention, the step of performing data scaling on the sensor charge array to obtain third array data, and extracting data feature information of the third array data through the activation function of the convolutional neural network specifically includes:

[0023] The sensor charge array is divided by a preset scaling factor to obtain a third array data, where the scaling factor is C k ×V LSB ,in, k Indicates the capacitance value of the preset equivalent capacitor structure, V LSB Indicates the preset analog-to-digital converter accuracy;

[0024] The data feature information of the third array data is extracted by a preset activation function, where the activation function is one of a Sigmoid function, a Tanh function, and a ReLU function.

[0025] Furthermore, in one embodiment of the present invention, the step of inputting the data feature information into the fully connected layer of the convolutional neural network to calculate an output vector, and then determining the recognition result of the sensor-sensed image based on the output vector, specifically includes:

[0026] Calculating a plurality of output vectors based on the data feature information and a preset second weight parameter;

[0027] Normalizing the output vector to obtain several classification probabilities;

[0028] An index value corresponding to the classification probability with the largest value is obtained, and a recognition result of the sensor-sensed image is determined according to the index value.

[0029] Furthermore, in one embodiment of the present invention, the sensor-induced image recognition method further includes the step of training a convolutional neural network, which specifically includes:

[0030] Determining a training data set, inputting the training data set into the convolutional neural network to obtain a prediction classification result;

[0031] Determining a loss value of the convolutional neural network based on the predicted classification result and the label of the training data set;

[0032] Updating the weight parameters of the convolutional layer and the fully connected layer through a back propagation algorithm according to the loss value;

[0033] When the loss value reaches a preset first threshold or the number of iterations reaches a preset second threshold, the training is stopped and a trained convolutional neural network is obtained.

[0034] In a second aspect, an embodiment of the present invention provides a sensor-induced image recognition system, comprising:

[0035] a noise adding module, configured to obtain first array data of an image sensed by a sensor, and add Gaussian noise to the first array data to obtain second array data;

[0036] a pixel capacitance array determination module, configured to perform functional relationship conversion on the second array data to obtain a pixel capacitance array, and then input the pixel capacitance array into a pre-trained convolutional neural network;

[0037] A convolution calculation module, configured to perform convolution calculation on the pixel capacitance array through a convolution layer of the convolutional neural network to obtain a sensor charge array;

[0038] a feature extraction module, configured to perform data scaling on the sensor charge array to obtain third array data, and extract data feature information of the third array data through an activation function of the convolutional neural network;

[0039] The recognition module is used to input the data feature information into the fully connected layer of the convolutional neural network to calculate an output vector, and then determine the recognition result of the sensor-sensed image based on the output vector.

[0040] In a third aspect, an embodiment of the present invention provides a sensor-induced image recognition device, comprising:

[0041] at least one processor;

[0042] at least one memory for storing at least one program;

[0043] When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned sensor-induced image recognition method.

[0044] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium storing a program executable by a processor, wherein the program executable by the processor is used to execute the above-mentioned sensor-induced image recognition method when executed by the processor.

[0045] The advantages and benefits of the present invention will be described in part in the following description and will become apparent from the following description or learned through practice of the present invention:

[0046] The embodiment of the present invention adds Gaussian noise to the first array data of the sensor-sensed image to obtain the second array data, then performs functional relationship conversion on the second array data to obtain the pixel capacitance array, then inputs the pixel capacitance array into the trained convolutional neural network, performs convolution calculation on the pixel capacitance array through the convolution layer to obtain the sensor charge array, then performs data scaling on the sensor charge array to obtain the third array data, extracts the data feature information of the third array data through the activation function, and finally calculates the output vector through the fully connected layer. The recognition result of the sensor-sensed image can be determined based on the output vector. The embodiment of the present invention utilizes the electrical characteristics of the capacitor and uses analog capacitance values ​​to participate in the calculation, which can effectively simplify circuit resources, speed up the calculation process in the neural network, and realize the rapid recognition of the sensor-sensed image. It is also beneficial to speed up the calculation speed, reduce data bandwidth and simplify hardware resources during hardware mapping, reduce the requirements for system storage space and computing power, and improve the recognition efficiency of the sensor-sensed image. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following introduction is made to the drawings required for use in the embodiments of the present invention. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative work.

[0048] Figure 1 A flowchart of a sensor-induced image recognition method provided by an embodiment of the present invention;

[0049] Figure 2 A schematic diagram of the training and recognition process of a convolutional neural network provided by an embodiment of the present invention;

[0050] Figure 3 A schematic diagram of a capacitive sensor provided by an embodiment of the present invention introducing parasitic capacitance due to external stimulation;

[0051] Figure 4 A schematic diagram of a calculation process of a pixel capacitance array provided by an embodiment of the present invention;

[0052] Figure 5 A schematic diagram of the convolution calculation process provided by an embodiment of the present invention;

[0053] Figure 6 A structural block diagram of a sensor-induced image recognition system provided by an embodiment of the present invention;

[0054] Figure 7 This is a structural block diagram of a sensor-induced image recognition device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and are not to be construed as limiting the present invention. The step numbers in the following embodiments are provided for ease of explanation only and do not limit the order of the steps. The order of execution of the steps in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0056] In the description of the present invention, "a plurality" means two or more. The terms "first" and "second" are used solely to distinguish technical features and are not to be construed as indicating or implying relative importance, or as implicitly indicating the number of the indicated technical features, or as implicitly indicating the order of the indicated technical features. Furthermore, unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art.

[0057] Capacitive pixel array sensors typically utilize external sensors to generate parasitic capacitance, causing the capacitance of each pixel to change. A characteristic circuit then measures the capacitance of each pixel, converts it into digital data, and transmits it to a processor as an input feature image. The processor then runs a series of neural network algorithms to generate the sensor's sensing and recognition results. Reading and converting each capacitance value into a digital value requires significant time, storing the capacitive array image data requires significant storage space, and performing image data recognition calculations involves numerous convolution operations.

[0058] The embodiment of the present invention uses a capacitive fingerprint sensor as an example to describe the algorithm and circuit mapping. The capacitive fingerprint sensor is composed of a capacitor array, each unit is called a pixel, and the initial capacitance value of each capacitor is C0. When the fingerprint is pressed on the sensor, the finger brings parasitic capacitance to each capacitor, where the size of the parasitic capacitance is related to the distance from the epithelial cell layer of the finger to the sensor, that is, the groove fingerprint information of the finger will bring parasitic capacitance information to the sensor. The parasitic capacitance is connected in series with the initial capacitance, thereby changing the capacitance value of each pixel. Figure 3 Figure 2 shows a schematic diagram of a capacitive sensor provided by an embodiment of the present invention that introduces parasitic capacitance in response to external stimuli. The wavy curve represents the skin of a finger, the capacitor Ci within the dashed box represents the parasitic capacitance introduced by the finger, and the capacitor C0 outside the dashed box represents the original capacitance. The capacitors form a series structure, and the non-series nodes of the parasitic capacitance are connected to the ground.

[0059] Reference Figure 1 The embodiment of the present invention provides a sensor-induced image recognition method, which specifically includes the following steps:

[0060] S101, obtaining first array data of a sensor-sensed image, and adding Gaussian noise to the first array data to obtain second array data;

[0061] Specifically, the first array data is two-dimensional array data, which is used to simulate and represent the parasitic capacitance introduced by the capacitance array sensor when sensing the outside world.

[0062] As an optional embodiment, the second array data is calculated by the following formula:

[0063] P(ij)=[1+σ(Noise)×randn(0,1)]Q(ij)

[0064] Wherein, P(ij) represents the data in the i-th row and j-th column of the second array data, Q(ij) represents the data in the i-th row and j-th column of the first array data, σ(Noise) represents the preset noise coefficient, and randn(0,1) represents a random number generated by a Gaussian function with a center of 0 and a variance of 1.

[0065] Specifically, Gaussian noise is added to the first array data by multiplying each data point of the first array data by [1+σ(Noise)×randn(0,1)]. This can simulate the noise existing when the capacitor array in the real circuit receives external information, so that the parasitic capacitance value has a slight noise fluctuation.

[0066] S102 , performing functional relationship conversion on the second array data to obtain a pixel capacitance array, and then inputting the pixel capacitance array into a pre-trained convolutional neural network.

[0067] Specifically, the second array data is converted using the capacitance series formula. The output pixel capacitance array corresponds to the total capacitance after the parasitic capacitance and the original capacitance are connected in series. When mapped to the circuit, the capacitance array sensor generates parasitic capacitance when receiving external information. The parasitic capacitance and the original capacitance form a series structure. Therefore, the capacitance series formula is applied to calculate the total pixel capacitance.

[0068] As a further optional implementation, the pixel capacitance array is calculated using the following formula:

[0069] C(ij)=[P(ij)×C0] / [P(ij)+C0]

[0070] Wherein, C(ij) represents the data of the i-th row and j-th column in the pixel capacitance array, P(ij) represents the data of the i-th row and j-th column in the second array data, and C0 represents the original capacitance of the sensor.

[0071] Specifically, each data in the calculated pixel capacitance array corresponds to the series capacitance value of the parasitic capacitance and the original capacitance. Figure 4A schematic diagram of a calculation process of a pixel capacitance array provided by an embodiment of the present invention is provided. Second array data is obtained by adding noise to first array data, and then data transformation is performed on the second array data to obtain a pixel capacitance array.

[0072] S103 , performing convolution calculation on the pixel capacitance array through the convolution layer of the convolutional neural network to obtain a sensor charge array.

[0073] Specifically, the input of the convolution layer is the pixel capacitance array obtained by converting the capacitor series formula. The convolution kernel weight corresponds to the voltage applied to the pixel capacitance array, and the output corresponds to the amount of charge on the pixel capacitance array. The convolution process is equivalent to applying a voltage to the capacitor and then obtaining the numerical value of the charge.

[0074] As a further optional implementation, the step of performing convolution calculation on the pixel capacitance array through the convolution layer of the convolutional neural network to obtain the sensor charge array is specifically as follows:

[0075] Performing convolution calculation on the pixel capacitance array according to a preset first weight parameter to obtain a sensor charge array;

[0076] The first weight parameter is used to represent the voltage value applied to the sensor pixel capacitance.

[0077] Specifically, applying voltage to the pixel capacitor can collect charge. Collecting the pixel charge into a unified circuit structure can obtain the total charge. In this embodiment of the present invention, the convolution layer performs a convolution calculation on the capacitance data using the voltage data as the weight. When this process is performed on m×n pixels, the charge amount obtained is:

[0078]

[0079] Wherein, V represents the first weight parameter, which can also be in array form.

[0080] The above formula can simulate the calculation of the convolution window, such as Figure 5 FIG2 is a schematic diagram of a convolution calculation process provided by an embodiment of the present invention, wherein a sensor charge array is obtained through convolution calculation.

[0081] S104 , performing data scaling on the sensor charge array to obtain third array data, and extracting data feature information of the third array data through an activation function of a convolutional neural network.

[0082] Specifically, the charge reading method can transfer the charge to a capacitor C_k of known capacity. At this time, the analog voltage across the capacitor is:

[0083] V A =Q / C_k

[0084] In order to read this voltage, an analog-to-digital converter is needed to convert the voltage into a digital value. The accuracy of the analog-to-digital converter is V LSB , the converted digital voltage data is:

[0085] V D =V A / V LSB

[0086] In the embodiment of the present invention, the conversion of digital voltage is achieved through data scaling, and then feature extraction is performed through an activation function.

[0087] Step S104 specifically includes the following steps:

[0088] S1041, divide the sensor charge array by a preset scaling factor to obtain third array data, where the scaling factor is C k ×V LSB , where C k Indicates the capacitance value of the preset equivalent capacitor structure, V LSB Indicates the preset analog-to-digital converter accuracy;

[0089] S1042. Extract data feature information of the third array data through a preset activation function, where the activation function is one of a Sigmoid function, a Tanh function, and a ReLU function.

[0090] Specifically, the sensor charge array is divided by the scaling factor and then output. The scaling factor is C k ×V LSB , where the left term corresponds to the capacitance value C k The equivalent capacitor structure, the charge divided by the capacitance corresponds to the voltage across the equivalent capacitor; the right-hand term corresponds to the use of the accuracy of V LSB The analog-to-digital converter reads the voltage across the equivalent capacitor to obtain binary data.

[0091] The resulting binary voltage data is passed to a digital processor for activation function calculation. This process eliminates analog computation and circuit parameter mapping. Instead, the activation function performs a nonlinear transformation on the extracted information to obtain the input data's features, or data feature information. The activation function can be any of the following: Sigmoid, Tanh, or ReLU functions.

[0092] S105: Input the data feature information into the fully connected layer of the convolutional neural network to calculate an output vector, and then determine the recognition result of the sensor-sensed image based on the output vector.

[0093] Specifically, the data feature information is classified and calculated through the fully connected layer, that is, the previously extracted data feature information is combined, and a corresponding output vector is output for each combination result. The output vector is used to represent the classification probability of each combination result. Step S105 specifically includes the following steps:

[0094] S1051. Calculate and obtain a plurality of output vectors based on the data feature information and a preset second weight parameter;

[0095] S1052. Normalize the output vector to obtain several classification probabilities;

[0096] S1053 , obtaining an index value corresponding to the classification probability with the largest value, and determining a recognition result of the sensor-sensed image according to the index value.

[0097] Specifically, the fully connected layer multiplies the input data feature information with the preset second weight parameter to obtain several output vectors corresponding to several classification results. This layer reorganizes the previously obtained data feature information; the output vector is normalized by the Softmax function to obtain the classification probability data with a value in the range of [0,1] and then output; the output classification probabilities are compared to find the index value corresponding to the maximum value, and the classification result is determined based on the index value, that is, the recognition result of the sensor-sensed image.

[0098] As an optional embodiment, the sensor-induced image recognition method further includes the step of training a convolutional neural network, which specifically includes:

[0099] A1. Determine the training data set and input it into the convolutional neural network to obtain the predicted classification results.

[0100] A2. Determine the loss value of the convolutional neural network based on the predicted classification results and the labels of the training dataset;

[0101] A3. Update the weight parameters of the convolutional layer and the fully connected layer through the back propagation algorithm according to the loss value;

[0102] A4. When the loss value reaches a preset first threshold or the number of iterations reaches a preset second threshold, training is stopped to obtain a trained convolutional neural network.

[0103] Specifically, through the aforementioned steps S101 to S105, a complete inference process of the convolutional neural network classification algorithm can be completed, and to implement the training process of the convolutional neural network, it is also necessary to calculate the loss value and update the weight parameters. Figure 2The figure shows a schematic diagram of the training and recognition process of the convolutional neural network provided by an embodiment of the present invention. In the embodiment of the present invention, the data processing process when training the convolutional neural network is similar to the data processing process during recognition, which will not be repeated here. The difference is that when training the model, the loss value is calculated for the obtained predicted classification results and the preset label data, and then the gradient of the loss value to the weight parameter is calculated using the original weight parameter. The weight parameter is updated while adjusting the learning rate, and is output to the convolution layer and the fully connected layer to replace the original weight parameter. The training is completed until the loss value reaches the preset first threshold or the number of iterations reaches the preset second threshold.

[0104] In an embodiment of the present invention, after the training data set is input into the initialized convolutional neural network, the predicted classification result output by the model can be obtained. The accuracy of the recognition model prediction can be evaluated based on the predicted classification result and the label of the training data set, thereby updating the parameters of the model. For a convolutional neural network, the accuracy of the model prediction result can be measured by a loss function. The loss function is defined on a single training data and is used to measure the prediction error of a training data. Specifically, the loss value of the training data is determined by the label of the single training data and the prediction result of the model for the training data. During actual training, a training data set has a lot of training data, so a cost function is generally used to measure the overall error of the training data set. The cost function is defined on the entire training data set and is used to calculate the average value of the prediction error of all training data, which can better measure the prediction effect of the model. For a general machine learning model, based on the aforementioned cost function, a regularization term that measures the complexity of the model can be used as the objective function of the training. Based on this objective function, the loss value of the entire training data set can be calculated. There are many types of commonly used loss functions, such as 0-1 loss function, square loss function, absolute loss function, logarithmic loss function, cross entropy loss function, etc., which can all be used as loss functions of machine learning models, which will not be elaborated here one by one. In an embodiment of the present invention, any one of the loss functions can be selected to determine the loss value of training. Based on the loss value of training, the back propagation algorithm is used to update the parameters of the model, and a trained convolutional neural network can be obtained by iterating several rounds. The specific number of iterations can be pre-set, or the training is considered to be completed when the test set meets the accuracy requirements.

[0105] The above describes the method steps of the present invention. It can be recognized that the embodiments of the present invention utilize the electrical characteristics of capacitors and use analog capacitance values ​​to participate in calculations, which can effectively simplify circuit resources, speed up the calculation process in the neural network, and achieve rapid recognition of sensor-sensed images. In addition, during hardware mapping, it is beneficial to speed up the calculation speed, reduce data bandwidth, and simplify hardware resources, thereby reducing the requirements for system storage space and computing power, and improving the recognition efficiency of sensor-sensed images.

[0106] Reference Figure 6 , an embodiment of the present invention provides a sensor-induced image recognition system, comprising:

[0107] a noise adding module, configured to obtain first array data of an image sensed by a sensor, and add Gaussian noise to the first array data to obtain second array data;

[0108] a pixel capacitance array determination module, configured to perform functional relationship conversion on the second array data to obtain a pixel capacitance array, and then input the pixel capacitance array into a pre-trained convolutional neural network;

[0109] A convolution calculation module is used to perform convolution calculation on the pixel capacitance array through the convolution layer of the convolutional neural network to obtain a sensor charge array;

[0110] a feature extraction module, configured to perform data scaling on the sensor charge array to obtain third array data, and extract data feature information of the third array data through an activation function of a convolutional neural network;

[0111] The recognition module is used to input data feature information into the fully connected layer of the convolutional neural network to calculate the output vector, and then determine the recognition result of the sensor-sensed image based on the output vector.

[0112] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0113] Reference Figure 7 , an embodiment of the present invention provides a sensor-induced image recognition device, comprising:

[0114] at least one processor;

[0115] at least one memory for storing at least one program;

[0116] When the at least one program is executed by the at least one processor, the at least one processor implements the sensor-induced image recognition method.

[0117] The contents of the above method embodiments are all applicable to the present device embodiments. The functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0118] An embodiment of the present invention further provides a computer-readable storage medium storing a program executable by a processor. When the program is executed by the processor, it is used to perform the above-mentioned sensor-induced image recognition method.

[0119] A computer-readable storage medium according to an embodiment of the present invention can execute a sensor-induced image recognition method provided by an embodiment of the method of the present invention, can execute any combination of implementation steps of the embodiment of the method, and has the corresponding functions and beneficial effects of the method.

[0120] The embodiment of the present invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs Figure 1 The method shown.

[0121] In some optional embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the above-mentioned boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided in an exemplary manner for the purpose of providing a more comprehensive understanding of the technology. The disclosed method is not limited to the operation and logic flow presented herein. Optional embodiments are contemplated in which the order of the various operations is changed and the sub-operations described as a part of a larger operation are performed independently.

[0122] In addition, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the above-mentioned functions and / or features can be integrated into a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It is also understood that a detailed discussion of the actual implementation of each module is not necessary for understanding the present invention. More specifically, given the properties, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the module will be understood within the routine skills of an engineer. Therefore, a person skilled in the art can implement the present invention set forth in the claims using ordinary skills without undue experimentation. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0123] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the above methods of each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0124] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0125] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable media on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0126] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0127] In the above description of this specification, reference to the terms "one embodiment / example," "another embodiment / example," or "certain embodiments / examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0128] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

[0129] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A sensor-induced image recognition method, characterized in that: The following steps are involved: Acquire first array data of a sensor-sensed image, and add Gaussian noise to the first array data to obtain second array data; Performing a functional relationship conversion on the second array data to obtain a pixel capacitance array, and then inputting the pixel capacitance array into a pre-trained convolutional neural network; Performing convolution calculation on the pixel capacitance array through the convolution layer of the convolutional neural network to obtain a sensor charge array; Performing data scaling on the sensor charge array to obtain third array data, and extracting data feature information of the third array data through the activation function of the convolutional neural network; Inputting the data feature information into the fully connected layer of the convolutional neural network to calculate an output vector, and then determining the recognition result of the sensor-sensed image based on the output vector; The second array data is calculated by the following formula: P(ij)=[1+σ(Noise)×randn(0,1)]Q(ij) Wherein, P(ij) represents the data in the i-th row and j-th column of the second array data, Q(ij) represents the data in the i-th row and j-th column of the first array data, σ(Noise) represents a preset noise coefficient, and randn(0, 1) represents a random number generated by a Gaussian function with a center of 0 and a variance of 1; The pixel capacitance array is calculated by the following formula: C(ij)=[P(ij)×C0] / [P(ij)+C0] Wherein, C(ij) represents the data of the i-th row and j-th column in the pixel capacitance array, P(ij) represents the data of the i-th row and j-th column in the second array data, and C0 represents the original capacitance of the sensor.

2. A sensor-induced image recognition method according to claim 1, characterized in that: The step of performing convolution calculation on the pixel capacitance array through the convolution layer of the convolutional neural network to obtain a sensor charge array is specifically as follows: Performing convolution calculation on the pixel capacitance array according to a preset first weight parameter to obtain the sensor charge array; The first weight parameter is used to represent the voltage value applied to the sensor pixel capacitance.

3. The sensor-induced image recognition method according to claim 1, characterized in that: The step of performing data scaling on the sensor charge array to obtain third array data, and extracting data feature information of the third array data through the activation function of the convolutional neural network specifically includes: The sensor charge array is divided by a preset scaling factor to obtain a third array data, where the scaling factor is C k ×V LSB , where C k Indicates the capacitance value of the preset equivalent capacitor structure, V LSB Indicates the preset analog-to-digital converter accuracy; The data feature information of the third array data is extracted by a preset activation function, where the activation function is one of a Sigmoid function, a Tanh function, and a ReLU function.

4. The sensor-induced image recognition method according to claim 1, wherein: The step of inputting the data feature information into the fully connected layer of the convolutional neural network to calculate an output vector, and then determining the recognition result of the sensor-sensed image according to the output vector specifically includes: Calculating a plurality of output vectors based on the data feature information and a preset second weight parameter; Normalizing the output vector to obtain several classification probabilities; An index value corresponding to the classification probability with the largest value is obtained, and a recognition result of the sensor-sensed image is determined according to the index value.

5. A sensor-induced image recognition method according to any one of claims 1 to 4, characterized in that: The sensor-induced image recognition method further includes the step of training a convolutional neural network, which specifically includes: Determining a training data set, inputting the training data set into the convolutional neural network to obtain a prediction classification result; Determining a loss value of the convolutional neural network based on the predicted classification result and the label of the training data set; Updating the weight parameters of the convolutional layer and the fully connected layer through a back propagation algorithm according to the loss value; When the loss value reaches a preset first threshold or the number of iterations reaches a preset second threshold, the training is stopped and a trained convolutional neural network is obtained.

6. A sensor-induced image recognition system, characterized in that: include: a noise adding module, configured to obtain first array data of an image sensed by a sensor, and add Gaussian noise to the first array data to obtain second array data; a pixel capacitance array determination module, configured to perform functional relationship conversion on the second array data to obtain a pixel capacitance array, and then input the pixel capacitance array into a pre-trained convolutional neural network; A convolution calculation module, configured to perform convolution calculation on the pixel capacitance array through a convolution layer of the convolutional neural network to obtain a sensor charge array; a feature extraction module, configured to perform data scaling on the sensor charge array to obtain third array data, and extract data feature information of the third array data through an activation function of the convolutional neural network; a recognition module, configured to input the data feature information into a fully connected layer of the convolutional neural network to calculate an output vector, and then determine a recognition result of the sensor-sensed image based on the output vector; The second array data is calculated by the following formula: P(ij)=[1+σ(Noise)×randn(0,1)]Q(ij) Wherein, P(ij) represents the data in the i-th row and j-th column of the second array data, Q(ij) represents the data in the i-th row and j-th column of the first array data, σ(Noise) represents a preset noise coefficient, and randn(0, 1) represents a random number generated by a Gaussian function with a center of 0 and a variance of 1; The pixel capacitance array is calculated by the following formula: C(ij)=[P(ij)×C0] / [P(ij)+C0] Wherein, C(ij) represents the data of the i-th row and j-th column in the pixel capacitance array, P(ij) represents the data of the i-th row and i-th column in the second array data, and C0 represents the original capacitance of the sensor.

7. A sensor-induced image recognition device, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the sensor-induced image recognition method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a program executable by a processor, characterized in that: The processor-executable program is used to perform a sensor-induced image recognition method according to any one of claims 1 to 5 when executed by the processor.

Images