Intelligent identification automobile data recorder based on thermal imaging and control method thereof

Abstract

The invention relates to the technical field of automobile data recorders, in particular to an intelligent identification automobile data recorder based on thermal imaging and a control method thereof.Through cooperation of a thermal imaging unit, a data processing unit, a target identification unit, a control unit, a communication unit and a man-machine interaction unit, the thermal imaging unit can receive infrared radiation emitted by a target object; the data processing unit processes and analyzes the electric signal to generate a clear thermal image, the target identification unit identifies and analyzes the thermal image to determine information such as the position, the shape and the size of a target object, and the control unit provides corresponding control signals according to the information. In this way, an image can be formed by detecting infrared radiation emitted by an object, imaging can be achieved without any external light source, clear views can be provided at night or under the low-light condition, and the safety of night driving is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of driving recorders, and in particular relates to an intelligent recognition driving recorder based on thermal imaging and a control method thereof. Background Art

[0002] The intelligent recognition driving recorder is an efficient device that integrates advanced intelligent technology. It captures video images during driving through a high-definition camera and uses built-in image processing chips and algorithms for real-time analysis. It can recognize license plates, traffic signs, pedestrians and other vehicles, etc. This recognition technology is based on deep learning and artificial intelligence, which can effectively improve road safety and reduce the occurrence of traffic accidents.

[0003] For example, the patent application number is 201910552168.6, which discloses a smart driving recorder, system and vehicle based on YOLO object recognition, including: a first camera, a second camera, a first chip, a second chip, a voice conversion chip, an image processor, a database server, a WIFI module, and a car audio system. First, the first camera and the second camera collect images, and then the first chip uses the YOLO object recognition algorithm to recognize the image processed by the image processor to obtain an image recognition result; the second chip performs distance measurement on the collected image to obtain a distance measurement result. The obtained object recognition result and distance measurement result are transmitted to the voice conversion chip, which processes them and outputs the voice to the car audio system, effectively improving the driving safety of the car owner and reducing the occurrence of accidents.

[0004] However, the above-mentioned smart driving recorder performs well during the day and under good lighting conditions, but at night or in a dimly lit environment, its accuracy and effect may be affected to a certain extent, thereby affecting the safety of drivers driving at night. Therefore, we need to propose an intelligent recognition driving recorder based on thermal imaging and its control method to solve the above-mentioned problems, so that it can form an image by detecting the infrared radiation emitted by the object, and can form an image without any external light source, so that it can provide a clear view at night or in low light conditions, thereby improving the safety of night driving. Summary of the invention

[0005] In view of the above problems, the present invention provides an intelligent recognition driving recorder based on thermal imaging, including a driving recorder body, wherein the driving recorder body is internally provided with a thermal imaging unit capable of clearly capturing the infrared radiation of a target object in a completely dark or extremely weak light environment, a data processing unit with a built-in AI deep learning algorithm, a target recognition unit with a long-sighted distance of 300 meters, a control unit for providing control signals, a communication unit capable of communication and positioning, and a human-computer interaction unit for performing human-computer interaction operations, the thermal imaging unit is electrically connected to the data processing unit, the data processing unit is electrically connected to the target recognition unit, the target recognition unit is electrically connected to the control unit, the control unit is electrically connected to the communication unit and the human-computer interaction unit, respectively, and the communication unit is electrically connected to the thermal imaging unit, the data processing unit and the target recognition unit, respectively.

[0006] Furthermore, the thermal imaging unit includes at least one infrared sensor, at least one of which is used to capture the infrared radiation signal of the target object, and then convert the captured infrared radiation signal into a digital signal through a signal processing circuit, and transmit the converted digital signal to the data processing unit for processing.

[0007] Furthermore, the signal processing circuit includes a coupling chip U1, a transmitting end of the coupling chip U1 is connected to a resistor R6, a receiving end of the coupling chip U1 is connected to an amplifier Q1, a positive end of the amplifier Q1 is connected to an adjustable resistor R8, one end of the adjustable resistor R8 is connected to a resistor R7, one end of the resistor R7 and the other end of the adjustable resistor R8 are respectively connected to two receiving ends of the coupling chip U1, and an output end of the coupling chip U1 is connected to a ground resistor R9.

[0008] Furthermore, the data processing unit includes a memory module for storing video data and at least one microprocessor for analyzing and processing images. The memory module is electrically connected to the microprocessor. After receiving the digital signal, the microprocessor analyzes and processes the image, and uses an AI deep learning algorithm to identify pedestrians, obstacles or vehicles in the image and respond accordingly.

[0009] Furthermore, the microprocessor performs the following process for analyzing and processing images:

[0010] B1. Extract the key features of the image through convolution operation. The convolution operation formula is as follows:

[0011]

[0012] Among them, f is the input signal, g is the convolution kernel, t is the time variable, τ is the virtual integral variable, and dτ is the small change of τ in the integral;

[0013] B2. Perform pooling operation based on the convolution operation result to calculate the sum of discrete convolution i jn , the calculation formula is as follows:

[0014]

[0015] Where m is the number of convolution kernels, x is the x-axis coordinate of the output area of ​​the convolution layer, and y is the y-axis coordinate of the output area of ​​the convolution layer;

[0016] B3. Find the maximum value of the output area after maximum pooling according to the following evaluation formula:

[0017] a x =maxi xy ,

[0018] Among them, i xy is a region of the convolutional layer output, a x is the maximum value of the region after pooling;

[0019] B4. Calculate the first layer weight value W1 according to the maximum output value. The calculation formula of the first layer weight value is as follows:

[0020] Among them, n is the weight matrix of the first layer;

[0021] B5. Calculate the second weight value W2. The calculation formula is as follows:

[0022] Among them, θ is the learning rate and L is the loss function;

[0023] B6. Calculate the third weight value W3. The calculation formula is as follows:

[0024]

[0025] B7. Use the output formula to classify and identify new data. The output formula is as follows:

[0026] Where k is the total number of categories.

[0027] Furthermore, the target identification unit includes the following steps when identifying the target:

[0028] A1. Preprocess the received image data using digital filters, including operations such as noise removal, contrast enhancement, and image resizing, in preparation for subsequent analysis;

[0029] A2. Extract key features from the preprocessed image. The key features include edges, textures, and shapes.

[0030] A3, dividing the image into multiple candidate regions so that the candidate regions can be processed and analyzed separately;

[0031] A4. Use a pre-trained convolutional neural network algorithm model to identify and locate objects in the image;

[0032] A5. Extract features into the convolutional neural network algorithm model for calculation, and output recognition results and target locations;

[0033] A6. Based on the identified target type and location, combined with scene analysis, determine whether the target has potential risks. If there is a potential risk, perform a warning operation. If there is no potential risk, proceed to the next image recognition.

[0034] Furthermore, the control unit includes a control circuit and an isolation circuit, and the control circuit is connected to the isolation circuit.

[0035] Furthermore, the control circuit includes a control chip U2, wherein pin 1 of the control chip U2 is connected to a capacitor C1, pin 2 of the control chip U2 is connected to a capacitor C2, and one end of the capacitor C2 and one end of the capacitor C1 are both connected to pin 5 of the control chip U2, pin 9 of the control chip U2 is connected to a resistor R61 and an operator U16 arranged in series, pin 11 of the control chip U2 is connected to a resistor R71 and an operator U17 arranged in series, pin 13 of the control chip U2 is connected to a resistor R81 and an operator U18 arranged in parallel, pin 10 of the operator U18 is connected to the other end of the capacitor C2, a resistor R13 is connected between pins 15 and 24 of the control chip U2, pin 21 of the control chip U2 is connected to a resistor R11, pin 19 of the control chip U2 is connected to a resistor R12, and one end of the resistor R11 and one end of the resistor R12 are both connected to pin 26 of the control chip U2.

[0036] Furthermore, the isolation circuit includes a coupling chip U21, a coupling chip U22, a coupling chip U23 and a coupling chip U24 which are arranged in parallel, the transmitting end of the coupling chip U21 is connected to a resistor R44, the transmitting end of the coupling chip U22 is connected to a resistor R43, the transmitting end of the coupling chip U23 is connected to a resistor R42, the transmitting end of the coupling chip U24 is connected to a resistor R41, and one end of the resistor R41, one end of the resistor R42, one end of the resistor R43 and one end of the resistor R44 are connected together, one of the receiving ends of the coupling chip U21 is connected to one of the receiving ends of the coupling chip U23, one of the receiving ends of the coupling chip U22 is connected to one of the receiving ends of the coupling chip U24, and a capacitor C55 is connected between the other receiving end of the coupling chip U23 and the other receiving end of the coupling chip U24.

[0037] Based on the above-described intelligent recognition driving recorder based on thermal imaging, the present invention also provides a control method of the intelligent recognition driving recorder based on thermal imaging, comprising the following steps:

[0038] S1, capturing the infrared radiation signal emitted by the target object through the thermal imaging unit and converting the infrared detection signal into a digital signal;

[0039] S2. Analyze and process the digital signal through the data processing unit, use the built-in AI deep learning algorithm to identify pedestrians, obstacles or vehicles in the image, and respond by triggering alarms or reminding attention according to the preset logic;

[0040] S3, preprocessing the digital signal through the target recognition unit, inputting the extracted features into the pre-trained convolutional neural network model, and finally outputting the recognition result and target position through multi-layer processing;

[0041] S4. According to the identified target type and position, combined with scene analysis, determine whether the target has potential risks. If there is a risk, perform a warning operation; if there is no risk, continue to monitor the next frame of the image;

[0042] S5. The user performs remote transmission, access and control with the human-computer interaction unit through the communication unit.

[0043] The beneficial effects of the present invention are:

[0044] 1. The present invention can form an image by detecting infrared radiation emitted by an object through the cooperation of a thermal imaging unit, a data processing unit, a target recognition unit, a control unit, a communication unit and a human-computer interaction unit. The image can be formed without any external light source, so that a clear view can be provided even at night or in low light conditions, thereby improving the safety of night driving.

[0045] 2. Through the setting of the data processing unit, the present invention can analyze and process the image after receiving the digital signal, use the AI ​​deep learning algorithm to identify the pedestrians, obstacles or vehicles in the image, and make corresponding responses, thereby improving the efficiency of data processing.

[0046] 3. The setting of the control unit of the present invention can manage and adjust the function of the entire circuit, ensure the normal operation of each unit, perform tasks according to a predetermined program, and establish an isolation barrier between circuits to ensure the safety and reliability of signal transmission and prevent safety problems caused by high voltage or faults.

[0047] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 A system block diagram according to an embodiment of the present invention is shown;

[0050] Figure 2 shows a circuit diagram of a signal processing circuit according to an embodiment of the present invention;

[0051] Figure 3 A circuit diagram of a control circuit according to an embodiment of the present invention is shown;

[0052] Figure 4 A circuit diagram of an isolation circuit according to an embodiment of the present invention is shown;

[0053] Figure 5 A flowchart according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] The embodiment of the present invention provides an intelligent recognition driving recorder based on thermal imaging, such as Figure 1-4As shown, it includes a driving recorder body, and the driving recorder body is internally provided with a thermal imaging unit capable of clearly capturing the infrared radiation of the target object in a completely dark or extremely weak light environment, a data processing unit with a built-in AI deep learning algorithm, a target recognition unit with a long-sighted distance of 300 meters, a control unit for providing control signals, a communication unit capable of communication and positioning, and a human-computer interaction unit for performing human-computer interaction operations, the thermal imaging unit is electrically connected to the data processing unit, the data processing unit is electrically connected to the target recognition unit, the target recognition unit is electrically connected to the control unit, the control unit is electrically connected to the communication unit and the human-computer interaction unit, respectively, and the communication unit is electrically connected to the thermal imaging unit, the data processing unit and the target recognition unit, respectively.

[0056] The thermal imaging unit includes at least one infrared sensor, which is used to capture the infrared radiation signal of the target object, and then convert the captured infrared radiation signal into a digital signal through a signal processing circuit, and transmit the converted digital signal to the data processing unit for processing. By capturing the infrared radiation emitted by the object to form an image, a clear view is provided for strengthening at night or in low light conditions, which significantly improves driving safety.

[0057] The signal processing circuit includes a coupling chip U1, a transmitting end of the coupling chip U1 is connected to a resistor R6, a receiving end of the coupling chip U1 is connected to an amplifier Q1, a positive end of the amplifier Q1 is connected to an adjustable resistor R8, one end of the adjustable resistor R8 is connected to a resistor R7, one end of the resistor R7 and the other end of the adjustable resistor R8 are respectively connected to two receiving ends of the coupling chip U1, and an output end of the coupling chip U1 is connected to a ground resistor R9. The energy of the input signal is effectively transferred to the next-level circuit through the coupling chip U1, and the input and output circuits are isolated to ensure that they do not interfere with each other, and effective energy transfer is achieved at the same time. The coupling chip U1 can efficiently transfer the energy of the infrared signal from the transmitting end to the receiving end, which is necessary for subsequent signal amplification and processing. It ensures that the signal will not be weakened due to energy loss during the transmission process. The amplifier Q1 realizes precise adjustment of the infrared signal gain by connecting the adjustable resistor R8 at its positive end. According to actual needs, the adjustable resistor R8 can be adjusted to change the amplification factor of the amplifier Q1 to adapt to different application scenarios.

[0058] The data processing unit includes a memory module for storing video data and at least one microprocessor for analyzing and processing images. The memory module is electrically connected to the microprocessor. After receiving the digital signal, the microprocessor analyzes and processes the image, uses an AI deep learning algorithm to identify pedestrians, obstacles or vehicles in the image, and performs corresponding responses, such as triggering an alarm. The cooperation between the storage module and the microprocessor provides the driving recorder with sufficient computing power and storage space to perform complex AI tasks;

[0059] The process of image analysis and processing by the microprocessor is as follows:

[0060] B1. Extract the key features of the image through convolution operation. The convolution operation formula is as follows:

[0061]

[0062] Among them, f is the input signal, g is the convolution kernel, t is the time variable, τ is the virtual integral variable, and dτ is the small change of τ in the integral;

[0063] B2. Perform pooling operation based on the convolution operation results to calculate the sum of discrete convolution i jn , the calculation formula is as follows:

[0064]

[0065] Where m is the number of convolution kernels, x is the x-axis coordinate of the output area of ​​the convolution layer, and y is the y-axis coordinate of the output area of ​​the convolution layer;

[0066] B3. Find the maximum value of the output area after maximum pooling according to the following evaluation formula:

[0067] a x =maxi xy ,

[0068] Among them, i xy is a region of the convolutional layer output, a x is the maximum value of the region after pooling;

[0069] B4. Calculate the first layer weight value W1 according to the maximum output value. The calculation formula of the first layer weight value is as follows:

[0070] Among them, n is the weight matrix of the first layer;

[0071] B5. Calculate the second weight value W2. The calculation formula is as follows:

[0072] Among them, θ is the learning rate and L is the loss function;

[0073] B6. Calculate the third weight value W3. The calculation formula is as follows:

[0074]

[0075] B7. Use the output formula to classify and identify new data. The output formula is as follows:

[0076] Where k is the total number of categories.

[0077] The memory module includes an internal memory and an external storage interface, the internal memory is used to store recorded video data, the external memory is used to store recorded video data, and the external storage interface is used for users to export data to other devices to improve the convenience of data transmission.

[0078] After receiving the data transmitted by the data processing unit, the target recognition unit further analyzes the image to accurately recognize the target and determine whether the detected target has potential risks.

[0079] The target identification unit comprises the following steps when identifying a target:

[0080] A1. Preprocess the received image data using digital filters, including operations such as noise removal, contrast enhancement, and image resizing, in preparation for subsequent analysis;

[0081] A2. Extract key features from the preprocessed image. The key features include edges, textures, and shapes.

[0082] A3, dividing the image into multiple candidate regions so that the candidate regions can be processed and analyzed separately;

[0083] A4. Use a pre-trained convolutional neural network algorithm model to identify and locate targets in images; the convolutional neural network algorithm model is a deep learning model that has achieved great success in the field of computer vision by simulating the way human visual processing is done. It achieves efficient recognition and classification of images through multiple layers of cascaded convolution, activation, pooling, and fully connected layers. These layers sequentially extract features, introduce nonlinearity, reduce data dimensions, and map features to outputs to complete complex image processing tasks. The convolutional neural network algorithm model uses common knowledge technology in the field and will not be described in detail here.

[0084] A5. Extract features into the convolutional neural network algorithm model for calculation, and output recognition results and target locations;

[0085] A6. According to the identified target type and location, combined with scene analysis, determine whether the target has potential risks. If there is a potential risk, perform a warning operation. If there is no potential risk, proceed to the next image recognition;

[0086] The control unit includes a control circuit and an isolation circuit. The control circuit is connected to the isolation circuit. The control circuit manages and adjusts the function of the entire circuit to ensure the normal operation of each unit and perform tasks according to a predetermined program. The isolation circuit is used to establish an isolation barrier between circuits to ensure the safety and reliability of signal transmission and prevent safety problems caused by high voltage or faults.

[0087] The control circuit includes a control chip U2, wherein the pin 1 of the control chip U2 is connected to a capacitor C1, the pin 2 of the control chip U2 is connected to a capacitor C2, and one end of the capacitor C2 and one end of the capacitor C1 are both connected to the pin 5 of the control chip U2, the pin 9 of the control chip U2 is connected to a resistor R61 and an operator U16 arranged in series, the pin 11 of the control chip U2 is connected to a resistor R71 and an operator U17 arranged in series, and the pin 13 of the control chip U2 is connected to a resistor R81 and an operator U17 arranged in parallel. U18, the 10th pin of the operator U18 is connected to the other end of the capacitor C2, a resistor R13 is connected between the 15th pin and the 24th pin of the control chip U2, the 21st pin of the control chip U2 is connected to the resistor R11, the 19th pin of the control chip U2 is connected to the resistor R12, and one end of the resistor R11 and one end of the resistor R12 are both connected to the 26th pin of the control chip U2, and three negative voltage feedback networks are established using the operator U16, the operator U17, and the operator U18 to stabilize the gain of the operator or improve its frequency response.

[0088] The isolation circuit includes a coupling chip U21, a coupling chip U22, a coupling chip U23 and a coupling chip U24 which are arranged in parallel. The transmitting end of the coupling chip U21 is connected to a resistor R44, the transmitting end of the coupling chip U22 is connected to a resistor R43, the transmitting end of the coupling chip U23 is connected to a resistor R42, the transmitting end of the coupling chip U24 is connected to a resistor R41, and one end of the resistor R41, one end of the resistor R42, one end of the resistor R43 and one end of the resistor R44 are connected together, one of the receiving ends of the coupling chip U21 is connected to one of the receiving ends of the coupling chip U23, one of the receiving ends of the coupling chip U22 is connected to one of the receiving ends of the coupling chip U24, and a capacitor C55 is connected between the other receiving end of the coupling chip U23 and the other receiving end of the coupling chip U24. Connecting multiple coupling chips in parallel can enhance the signal processing capability. In the parallel configuration, each coupling chip can work independently to disperse the load and improve the efficiency and reliability of the overall circuit. The coupling chips connected in parallel can achieve signal synchronization through a common collector or emitter connection point (i.e., the common connection point of all resistors R41, R42, R43, and R44), thereby minimizing the transmission delay of the signal on each coupling path and optimizing the overall transmission speed and quality of the signal.

[0089] The communication unit includes one or more of a Wi-Fi module, a Bluetooth module and a Beidou module. The Wi-Fi module, the Bluetooth module and the Beidou module are all existing technologies. Through the communication unit, the user can remotely access the data, configuration and other related information of the driving recorder, thereby improving the signal transmission between the units.

[0090] The human-computer interaction unit includes a physical button, a touch screen and a voice controller. The physical button realizes interactive operation by closing a circuit, that is, when the user presses the physical button, the physical button closes a circuit, allowing current to pass, so that the driving recorder sends a signal; the touch screen uses a resistive or capacitive sensing method to detect the position touched by the user. When the user's finger touches the screen, it causes a change in capacitance, thereby accurately measuring the touch position;

[0091] The voice controller includes an acoustic model, a language model and a dictionary, which converts the user's voice signal into text. The acoustic model is used to identify sound features, the language model is used to understand language patterns, and the dictionary is used to identify specific words and phrases. The user only needs to issue a command sound to perform the corresponding operation, thereby improving the convenience of user use.

[0092] Through the cooperation of the thermal imaging unit, the data processing unit, the target recognition unit, the control unit, the communication unit and the human-computer interaction unit, the thermal imaging unit will receive the infrared radiation emitted by the target object and convert it into electrical signals, which will be transmitted to the data processing unit. The data processing unit with built-in AI deep learning algorithm will process and analyze the electrical signals to generate clear thermal images. Then, the thermal images will be identified and analyzed by the target recognition unit to determine the position, shape, size and other information of the target object. This information will be transmitted to the control unit, and the control unit will provide corresponding control signals based on this information. At the same time, the communication unit can communicate and locate, and transmit the location information of the driving recorder to other devices or systems. The human-computer interaction unit is used to perform human-computer interaction operations, such as adjusting parameters and viewing images. In this way, images can be formed by detecting the infrared radiation emitted by the object, and imaging can be achieved without any external light source, so that a clear view can be provided at night or in low light conditions, thereby improving the safety of night driving.

[0093] Based on the above-described intelligent recognition driving recorder based on thermal imaging, the present invention also provides a control method for the intelligent recognition driving recorder based on thermal imaging, such as Figure 5 As shown, the following steps are included:

[0094] S1, capturing the infrared radiation signal emitted by the target object through the thermal imaging unit and converting the infrared detection signal into a digital signal;

[0095] S2. Analyze and process the digital signal through the data processing unit, use the built-in AI deep learning algorithm to identify pedestrians, obstacles or vehicles in the image, and respond by triggering alarms or reminding attention according to the preset logic;

[0096] S3, preprocessing the digital signal through the target recognition unit, inputting the extracted features into the pre-trained convolutional neural network model, and finally outputting the recognition result and target position through multi-layer processing;

[0097] S4. According to the identified target type and position, combined with scene analysis, determine whether the target has potential risks. If there is a risk, perform a warning operation; if there is no risk, continue to monitor the next frame of the image;

[0098] S5. Users can remotely transmit, access and control the human-computer interaction unit through the communication unit to improve the convenience of use.

[0099] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent recognition driving recorder based on thermal imaging, characterized in that: The driving recorder comprises a driving recorder body, wherein a thermal imaging unit capable of clearly capturing infrared radiation of a target object in a completely dark or extremely weak light environment, a data processing unit with a built-in AI deep learning algorithm, a target recognition unit with a long-sightedness of 300 meters, a control unit for providing control signals, a communication unit capable of communication and positioning, and a human-computer interaction unit for performing human-computer interaction operations are arranged inside the driving recorder body, wherein the thermal imaging unit is electrically connected to the data processing unit, the data processing unit is electrically connected to the target recognition unit, the target recognition unit is electrically connected to the control unit, the control unit is electrically connected to the communication unit and the human-computer interaction unit respectively, and the communication unit is electrically connected to the thermal imaging unit, the data processing unit and the target recognition unit respectively.

2. The intelligent recognition driving recorder based on thermal imaging according to claim 1, characterized in that: The thermal imaging unit includes at least one infrared sensor, which is used to capture the infrared radiation signal of the target object, convert the captured infrared radiation signal into a digital signal through a signal processing circuit, and transmit the converted digital signal to the data processing unit for processing.

3. The intelligent recognition driving recorder based on thermal imaging according to claim 2, characterized in that: The signal processing circuit includes a coupling chip U1, a transmitting end of the coupling chip U1 is connected to a resistor R6, a receiving end of the coupling chip U1 is connected to an amplifier Q1, a positive end of the amplifier Q1 is connected to an adjustable resistor R8, one end of the adjustable resistor R8 is connected to a resistor R7, one end of the resistor R7 and the other end of the adjustable resistor R8 are respectively connected to two receiving ends of the coupling chip U1, and an output end of the coupling chip U1 is connected to a ground resistor R9.

4. The intelligent recognition driving recorder based on thermal imaging according to claim 3 is characterized in that: The data processing unit includes a memory module for storing video data and at least one microprocessor for analyzing and processing images. The memory module is electrically connected to the microprocessor. After receiving the digital signal, the microprocessor analyzes and processes the image, and uses an AI deep learning algorithm to identify pedestrians, obstacles or vehicles in the image and respond accordingly.

5. The intelligent recognition driving recorder based on thermal imaging according to claim 4, characterized in that: The process of image analysis and processing by the microprocessor is as follows: B1. Extract the key features of the image through convolution operation. The convolution operation formula is as follows: Among them, f is the input signal, g is the convolution kernel, t is the time variable, τ is the virtual integral variable, and dτ is the small change of τ in the integral; B2. Perform pooling operation based on the convolution operation result to calculate the sum of discrete convolution i jn , the calculation formula is as follows: Where m is the number of convolution kernels, x is the x-axis coordinate of the output area of ​​the convolution layer, and y is the y-axis coordinate of the output area of ​​the convolution layer; B3. Find the maximum value of the output area after maximum pooling according to the following evaluation formula: to x maxi xy , Among them, i xy is a region of the convolutional layer output, a x is the maximum value of the region after pooling; B4. Calculate the first layer weight value W1 according to the maximum output value. The calculation formula of the first layer weight value is as follows: Among them, n is the weight matrix of the first layer; B5. Calculate the second weight value W2. The calculation formula is as follows: Among them, θ is the learning rate and L is the loss function; B6. Calculate the third weight value W3. The calculation formula is as follows: B7. Use the output formula to classify and identify new data. The output formula is as follows: Where k is the total number of categories.

6. The intelligent recognition driving recorder based on thermal imaging according to claim 5, characterized in that: The target identification unit comprises the following steps when identifying a target: A1. Preprocess the received image data using digital filters, including operations such as noise removal, contrast enhancement, and image resizing, in preparation for subsequent analysis; A2. Extract key features from the preprocessed image. The key features include edges, textures, and shapes. A3, dividing the image into multiple candidate regions so that the candidate regions can be processed and analyzed separately; A4. Use a pre-trained convolutional neural network algorithm model to identify and locate objects in the image; A5. Extract features into the convolutional neural network algorithm model for calculation, and output recognition results and target locations; A6. Based on the identified target type and location, combined with scene analysis, determine whether the target has potential risks. If there is a potential risk, perform a warning operation. If there is no potential risk, proceed to the next image recognition.

7. The intelligent recognition driving recorder based on thermal imaging according to claim 6, characterized in that: The control unit comprises a control circuit and an isolation circuit, and the control circuit is connected to the isolation circuit.

8. The intelligent recognition driving recorder based on thermal imaging according to claim 7, characterized in that: The control circuit includes a control chip U2, wherein pin 1 of the control chip U2 is connected to a capacitor C1, pin 2 of the control chip U2 is connected to a capacitor C2, and one end of the capacitor C2 and one end of the capacitor C1 are both connected to pin 5 of the control chip U2, pin 9 of the control chip U2 is connected to a resistor R61 and an operator U16 arranged in series, pin 11 of the control chip U2 is connected to a resistor R71 and an operator U17 arranged in series, pin 13 of the control chip U2 is connected to a resistor R81 and an operator U18 arranged in parallel, pin 10 of the operator U18 is connected to the other end of the capacitor C2, a resistor R13 is connected between pins 15 and 24 of the control chip U2, pin 21 of the control chip U2 is connected to a resistor R11, pin 19 of the control chip U2 is connected to a resistor R12, and one end of the resistor R11 and one end of the resistor R12 are both connected to pin 26 of the control chip U2.

9. The intelligent recognition driving recorder based on thermal imaging according to claim 8, characterized in that: The isolation circuit includes a coupling chip U21, a coupling chip U22, a coupling chip U23 and a coupling chip U24 which are arranged in parallel. The transmitting end of the coupling chip U21 is connected to a resistor R44, the transmitting end of the coupling chip U22 is connected to a resistor R43, the transmitting end of the coupling chip U23 is connected to a resistor R42, the transmitting end of the coupling chip U24 is connected to a resistor R41, and one end of the resistor R41, one end of the resistor R42, one end of the resistor R43 and one end of the resistor R44 are connected together, one of the receiving ends of the coupling chip U21 is connected to one of the receiving ends of the coupling chip U23, one of the receiving ends of the coupling chip U22 is connected to one of the receiving ends of the coupling chip U24, and a capacitor C55 is connected between the other receiving end of the coupling chip U23 and the other receiving end of the coupling chip U24.

10. A control method for a smart identification driving recorder based on thermal imaging, based on the smart identification driving recorder based on thermal imaging according to any one of claims 1 to 9, characterized in that: The steps include: S1, capturing the infrared radiation signal emitted by the target object through the thermal imaging unit and converting the infrared detection signal into a digital signal; S2. Analyze and process the digital signal through the data processing unit, use the built-in AI deep learning algorithm to identify pedestrians, obstacles or vehicles in the image, and respond by triggering alarms or reminding attention according to the preset logic; S3, preprocessing the digital signal through the target recognition unit, inputting the extracted features into the pre-trained convolutional neural network model, and finally outputting the recognition result and target position through multi-layer processing; S4. Based on the identified target type and location, combined with scene analysis, determine whether the target has potential risks. If there is a risk, perform warning operations; If there is no risk, continue to monitor the next frame of image; S5. The user performs remote transmission, access and control with the human-computer interaction unit through the communication unit.

Images