Intelligent code reading equipment based on AI

Abstract

The invention provides intelligent code reading equipment based on AI, and relates to the technical field of Internet. The equipment is provided with a shell which is made of an aluminum alloy material and is used for protecting internal vulnerable components; a light source module is arranged at the front end of the device and internally provided with multiple LED lamps with different wavelengths, and the illumination mode can be actively adjusted according to the actual environment requirement. A high-definition CMOS image sensor is used as an image acquisition module to complete image acquisition of a target area; the image processing module is integrated with a high-performance DSP chip, analyzes the original image acquired by the image acquisition module based on an AI algorithm, and outputs a decoding result; the environment sensor detects environment light information in real time and is used for optimizing light source brightness; the control module integrates the functions of each internal module, can command the brightness of the LED lamp according to environmental data, controls the image acquisition and processing process, and finally achieves the quick and accurate decoding of the identification code. According to the scheme, the equipment can intelligently adapt to environment change, and the image quality is improved, so that effective decoding is realized.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to an AI-based intelligent code reading device. Background Art

[0002] In today's digital age, barcodes and QR codes are widely used in various fields, such as logistics, retail, manufacturing, etc., for product identification, information tracing, inventory management, etc. As an important tool for reading barcode and QR code information, the performance and accuracy of code reading equipment are crucial to improving work efficiency and management level.

[0003] At present, barcode reading equipment generally relies on laser scanning and image processing technology to work. When light shines on the barcode, the black part absorbs the light and the white part reflects the light. The optical sensor can detect the difference in the reflected light. Laser scanning technology emits a laser beam to the surface of the barcode, receives the reflected light signal, and then decodes it.

[0004] However, in actual applications, strong light environments will cause reflections and glare, affecting the sensor's signal reception. In low light environments, the signal is weak and prone to recognition errors. Therefore, existing methods are difficult to quickly decode under unstable light conditions. Summary of the invention

[0005] The present application provides an AI-based intelligent code reading device, which is used to enable the code reading device to obtain images of good quality under different ambient lighting conditions, reduce the impact of changes in lighting conditions on reading accuracy, and thus improve the overall code reading success rate of the device.

[0006] In the first aspect, the present application provides an AI-based intelligent code reading device, comprising: a shell made of aluminum alloy for protecting internal components; a light source module, arranged at the front end of the device, with multiple LED lights of different wavelengths built in, the light source module is used to adjust the lighting mode according to actual needs; an image acquisition module, using a high-definition CMOS sensor, for acquiring original image data of a target area; an image processing module, communicating with the image acquisition module, the image processing module integrating a high-performance DSP chip, the image processing module is used to analyze the image data in combination with an AI recognition algorithm after receiving the original image data transmitted by the image acquisition module to obtain a decoding result; an environmental sensor, arranged outside the shell, for determining ambient light data; a control module, communicating with the light source module, the image processing module and the environmental sensor, when the control module receives a device start-up instruction, determines the ambient light data through the environmental sensor, and then controls the brightness of the LED light according to the ambient light data, and after a set time, controls the image acquisition module to shoot the target area at a set frequency to obtain original image data, and sends the original image data to the image processing module to obtain a decoding result.

[0007] By adopting the above technical solution, the AI ​​code reading device can intelligently adjust the light source brightness and image acquisition parameters according to different ambient light conditions, thereby improving the image quality, improving the recognition accuracy of the image processing algorithm, and thus improving the overall code reading success rate.

[0008] In combination with some embodiments of the first aspect, in some embodiments, it also includes a shooting angle determination module, which is arranged in the image acquisition module and is used to detect the shooting angle of the image acquisition module in real time; a shooting distance determination module is arranged at the front end of the device and is used to determine the shooting distance between the device and the target area.

[0009] By adopting the above technical solution, the shooting parameters during image acquisition can be obtained in real time, and these parameters reflect the relative position relationship between the device and the target. The data obtained from these parameters can provide a basis for subsequent image correction, so that the image preprocessing unit can correct the original image in terms of angle, distance, etc. according to the changes in shooting conditions, compensate for the impact of shooting parameters on image quality, improve the model's ability to recognize images, and achieve a higher recognition accuracy.

[0010] In combination with some embodiments of the first aspect, in some embodiments, a deep learning model building module is further included, which is communicatively connected to the control module. The model building module is used to train a deep learning model based on the image samples and the corresponding shooting conditions after collecting image samples under multiple different shooting conditions. The shooting conditions include different ambient brightness, image acquisition angles and distances; the control module is communicatively connected to the shooting angle determination module and the shooting distance determination module, and the control module is also used to input the acquired shooting angle, shooting distance, ambient light data and original image data into the model building module to obtain the original image data after image correction under different shooting conditions.

[0011] By adopting the above technical solution, the deep learning model can be assisted to perform image correction in real time according to changes in shooting angle and distance, so that the image presents a more standardized state. This can reduce the recognition difficulty caused by different shooting conditions and improve the generalization performance of the model, thereby obtaining a higher successful recognition rate.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the image processing module includes a decoding model determination unit, which includes multiple decoding models and is used to determine different decoding models for decoding according to the type and characteristics of the identification code.

[0013] By adopting the above technical solution, the optimal decoding model can be selected for identification codes of different types and characteristics to avoid misidentification and missed identification of the model, improve the accuracy of decoding, and thus enhance the adaptability of the device.

[0014] In combination with some embodiments of the first aspect, in some embodiments, the image processing module includes an image preprocessing unit, an image segmentation unit, a feature extraction unit and an identification code decoding unit, wherein the image preprocessing unit is used to preprocess the collected original image data to obtain processed image data; the image segmentation unit is used to extract an image area containing an identification code from the processed image data using an image processing algorithm; the feature extraction unit is used to extract feature information of the identification code from the image area containing the identification code; and the identification code decoding unit is used to use a corresponding decoding model based on the feature information to obtain information contained in the identification code.

[0015] By adopting the above technical solution, the AI ​​code reading device can effectively extract the target area, obtain key features, and finally accurately decode through the image processing pipeline, making the entire process efficient and smooth, and improving the real-time response capability of the device.

[0016] In combination with some embodiments of the first aspect, in some embodiments, a display module is further included, which is used to display the code reading result and the working status of the device to the user.

[0017] By adopting the above technical solution, the AI ​​code reading device can intuitively present the recognition results and working status to the user, making it easier for the user to understand and judge the usage of the device, thereby improving the user experience.

[0018] In combination with some embodiments of the first aspect, in some embodiments, the light source module includes an LED lamp rotation unit and a reflection compensation unit, the LED lamp rotation unit is used to adjust the illumination angle of the LED lamp; the reflection compensation unit is communicatively connected to the environmental sensor, and the reflection compensation unit is used to monitor the reflection intensity of the target area in real time through the environmental sensor, and when the reflection intensity is greater than a set threshold, the wavelength, intensity and illumination angle of the LED lamp are adjusted to compensate for the reflection.

[0019] By adopting the above technical solution, the AI ​​code reading device can actively adjust the reflection conditions of the target area so that the image quality is not affected by the reflection, avoiding recognition failure caused by reflection and improving the device's ability to adapt to strong lighting environments.

[0020] In combination with some embodiments of the first aspect, in some embodiments, a power supply module is further included to provide power to various components in the device.

[0021] By adopting the above technical solution, the AI ​​code reading device provides stable working power to internal components through the power supply module, ensuring the reliable operation of the equipment.

[0022] In combination with some embodiments of the first aspect, in some embodiments, a network module is also included to realize the connection between each module in the device and the network. The network module is connected to the control module. After the code reading is completed, the network module uploads the code reading result to the set cloud.

[0023] By adopting the above technical solution, the AI ​​code reading device can realize cloud storage of recognition results, facilitate subsequent data query and analysis, and expand the application scope of the device.

[0024] In combination with some embodiments of the first aspect, in some embodiments, an alarm module is further included, which is communicatively connected to the control module and is used to issue a sound or light alarm reminder after receiving the code reading failure information sent by the control module.

[0025] By adopting the above technical solution, an alarm can be issued to prompt the user of identification failure, so that the user can understand the working status of the device in time, thereby improving the availability of the device.

[0026] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Due to the technical means of adjusting the light source and image acquisition parameters in real time according to the ambient light, the problem of image quality degradation caused by changes in lighting conditions is effectively solved, thereby achieving the technical effect of ensuring image quality under different lighting conditions and improving the success rate of reading.

[0027] 2. Due to the adoption of technical means to select the optimal decoding model for different identification code types and features, the problem of low reading accuracy of a single model is effectively solved, thereby achieving the technical effect of improving reading accuracy and adaptability.

[0028] 3. Due to the technical means of actively adjusting the light source according to the reflection situation in the target area, the problem of reading failure caused by reflection is effectively solved, thereby achieving the technical effect of improving the equipment's adaptability to strong light environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a module connection diagram of an embodiment of the present application; Figure 2 This is another module connection diagram of an embodiment of the present application.

[0030] Description of reference numerals: 1. Shell; 2. Light source module; 3. Image acquisition module; 4. Image processing module; 5. Environmental sensor; 6. Control module; 7. Shooting angle determination module; 8. Shooting distance determination module; 9. Deep learning model building module; 10. Decoding model determination unit; 11. Image preprocessing unit; 12. Image segmentation unit; 13. Feature extraction unit; 14. Identification code decoding unit; 15. Display module; 16. LED light rotation unit; 17. Reflection compensation unit; 18. Power supply module; 19. Network module; 20. Alarm module. DETAILED DESCRIPTION

[0031] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification and appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to and includes any or all possible combinations of one or more listed items.

[0032] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0033] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0034] The present application embodiment provides an AI-based intelligent code reading device, such as Figure 1 As shown, Figure 1 It is a module connection diagram of an embodiment of the present application, including: The housing 1 is made of aluminum alloy and is used to protect the internal components; The light source module is set at the front end of the device and has multiple LED lights with different wavelengths built in. The light source module is used to adjust the lighting mode according to actual needs; Image acquisition module 3, using a high-definition CMOS sensor, is used to collect raw image data of the target area; An image processing module 4 is connected to the image acquisition module 3 in communication. The image processing module 4 integrates a high-performance DSP chip. The image processing module 4 is used to analyze the image data in combination with an AI recognition algorithm to obtain a decoding result after receiving the original image data transmitted by the image acquisition module 3; An environmental sensor 5, arranged outside the housing 1, for determining ambient light data; The control module 6 is communicatively connected with the light source module, the image processing module 4 and the environmental sensor 5. When the control module 6 receives the device start-up instruction, it determines the ambient light data through the environmental sensor 5, and then controls the brightness of the LED lamp according to the ambient light data. After a set time, the image acquisition module 3 is controlled to shoot the target area at a set frequency to obtain original image data, and the original image data is sent to the image processing module 4 to obtain a decoding result.

[0035] Specifically, the housing 1 of the device is made of aluminum alloy, which is based on the excellent physical properties of aluminum alloy. Aluminum alloy has high strength and hardness, and can effectively resist external physical impact. For example, when the device accidentally falls or is hit, the housing 1 can withstand the impact force to prevent the internal precision components from being damaged. At the same time, the corrosion resistance of aluminum alloy also provides a guarantee for the long-term use of the device in a complex environment. In some environments where chemical substances may volatilize or the humidity is high, the housing 1 can avoid corrosion and maintain the integrity of its structure, thereby continuously and stably protecting the internal components and ensuring the normal operation of the device. The light source module is located at the front end of the device and is an important link in achieving high-quality image acquisition. It has built-in LED lights of multiple different wavelengths, and realizes the function of adjusting the lighting mode according to actual needs through a complex control mechanism. In actual operation, the device will determine the required lighting mode according to different code reading scenes and the characteristics of the target area. This process involves the analysis of multiple factors, such as the color, material and ambient light conditions of the target area. When facing a dark and light-absorbing target, a combination mode of LED lights with a longer wavelength and higher brightness may be selected to enhance the contrast of the target area so that the image acquisition module 3 can obtain a clearer image. For targets with strong reflective surfaces, the wavelength and intensity of the LED light will be adjusted to avoid excessive reflected light affecting image quality. This method of dynamically adjusting the lighting mode according to actual conditions is achieved through an intelligent algorithm built into the control module 6, which comprehensively considers the data fed back by various sensors to make the best lighting mode selection.

[0036] The image acquisition module 3 uses a high-definition CMOS sensor, which is an advanced technology currently widely used in the field of image acquisition. The CMOS sensor has the advantages of high resolution, low power consumption and fast response. When working, it converts the light signal into an electrical signal by sensing the light in the target area, and then generates the original image data. Its high-resolution characteristics enable the collected image to contain rich detail information, whether it is a fine barcode or a complex two-dimensional code, it can be clearly captured. In order to ensure the quality of the collected image data, the image acquisition module 3 works in conjunction with the control module 6. The control module 6 accurately controls the work of the image acquisition module 3 according to pre-set parameters, such as shooting frequency and shooting timing. For example, after the light in the target area is stable and reaches the optimal lighting condition, the control module 6 will start the image acquisition module 3 to shoot at the set frequency to obtain the most accurate original image data. The image processing module 4 is connected to the image acquisition module 3 in communication and is the core processing part of the entire device. It integrates a high-performance DSP chip, which has a powerful digital signal processing capability and can quickly process a large amount of image data. After receiving the original image data transmitted by the image acquisition module 3, the image processing module 4 begins to play its key role. It uses advanced AI recognition algorithms to analyze image data. The AI ​​recognition algorithm is trained with a large number of image samples and can recognize various patterns and features in images. For example, for barcodes, the algorithm can accurately identify key information such as the width ratio and position relationship of bars and spaces; for QR codes, it can parse out the encoding rules and data content. During the analysis process, the DSP chip will quickly calculate the image data, including a series of operations such as image enhancement, noise removal, and feature extraction, to provide better data for the AI ​​recognition algorithm, thereby improving the accuracy of the decoding results.

[0037] The environment sensor 5 is arranged outside the housing 1, and its main function is to determine the ambient light data. It can sense the light intensity, color temperature and other parameters of the surrounding environment in real time through high-precision light sensing elements. These data are essential for the normal operation of the entire device. The environment sensor 5 transmits the collected ambient light data to the control module 6, and the control module 6 adjusts the brightness of the LED lamp in the light source module according to these data. For example, in a strong light environment, the environment sensor 5 detects high-brightness light, and the control module 6 will reduce the brightness of the LED lamp accordingly to avoid overexposure of the image; while in a dim environment, the brightness of the LED lamp will be increased to ensure that the target area has sufficient lighting. This adaptive adjustment mechanism based on ambient light data enables the device to obtain high-quality original image data in various lighting environments. The control module 6 is the central nervous system of the entire intelligent code reading device. It maintains communication connection with the light source module, the image processing module 4 and the environment sensor 5. When receiving the device start-up instruction from the user to start using the device, the control module 6 first starts the environment sensor 5 to obtain the ambient light data. The device start-up instruction can be an instruction automatically generated after the user presses the start button or touches the switch to start use. Based on these data, the control module 6 adjusts the brightness of the LED lamp through a complex control algorithm to optimize the lighting mode. After the set time has passed and the lighting conditions are stable, the control module 6 starts the image acquisition module 3 to shoot the target area at the set frequency. Subsequently, the acquired raw image data is sent to the image processing module 4 for processing. During the whole process, the control module 6 needs to accurately coordinate the working order and time intervals between the modules to ensure that each link can proceed smoothly. At the same time, the control module 6 also has certain fault tolerance and adaptive capabilities. For example, when a module has a short-term failure or data transmission is abnormal, it can take timely measures to make adjustments to ensure the overall stability and reliability of the equipment.

[0038] In summary, this AI-based intelligent code reading device achieves accurate and fast reading of various types of identification codes under complex ambient lighting through close collaboration between various modules and the application of advanced technologies.

[0039] For ease of understanding, the following further describes the modules of the system provided by this implementation in combination with the above embodiments. Figure 2 , Figure 2 This is another module connection diagram of an embodiment of the present application.

[0040] exist Figure 2In the embodiment, the AI-based intelligent code reading device further includes a shooting angle determination module 7, which is arranged in the image acquisition module 3 and used to detect the shooting angle of the image acquisition module 3 in real time; and a shooting distance determination module 8, which is arranged at the front end of the device and used to determine the shooting distance between the device and the target area. Specifically, the shooting angle determination module 7 is arranged in the image acquisition module 3, and its main function is to detect the shooting angle of the image acquisition module 3 in real time. The shooting angle determination module 7 can be equipped with high-precision inertial measurement sensors such as gyroscopes and accelerometers, which can sense the posture changes of the image acquisition module 3 in three-dimensional space. For example, when the device has a certain tilt during installation, or when the angle changes due to external factors (such as slight vibration, human movement, etc.) during use, the gyroscope can detect the rotation angular velocity around each coordinate axis, and the accelerometer can measure the acceleration changes of the module in each direction. By real-time acquisition and processing of these data, the module can obtain the current angle information of the image acquisition module 3. At the same time, the data collected by the inertial measurement sensor can also be filtered. Since the sensor may be interfered by noise during operation, such as electromagnetic interference of the device itself, slight vibration in the environment, etc., the filtering algorithm can effectively remove these noises and improve the accuracy of the data. Then, the coordinate transformation algorithm is used to convert the sensor data into a shooting angle relative to the target area. For example, the plane where the target area is located is used as the reference plane, and the shooting angle is accurately determined by calculating the angle between the image acquisition module 3 and the normal direction of the plane and the rotation angle in the plane. This function of real-time detection of the shooting angle is of great significance. In the actual code reading scene, different shooting angles may cause the shape, proportion and clarity of the target code in the image to change. When the shooting angle determination module 7 detects the angle change, it can feed back the data to the control module 6. The control module 6 adjusts the working parameters of other related modules based on this information. For example, if the shooting angle deviates from the preset optimal angle range, the control module 6 can adjust the irradiation direction of the LED light in the light source module to better illuminate the target area and reduce the shadow or reflection caused by the angle problem. At the same time, for the image processing module 4, the shooting angle information helps to perform geometric correction in the subsequent image analysis. If the image is deformed due to angle, the image processing algorithm can perform a reverse transformation based on the shooting angle data and restore the image to a state of approximate orthographic projection, thereby improving the AI ​​recognition algorithm's recognition accuracy for the code.

[0041] The shooting distance determination module 8 is located at the front end of the device, and its core task is to determine the shooting distance between the device and the target area. The front end of the device can be equipped with a special laser distance sensor or use the optical characteristics of the image acquisition module 3 itself for distance measurement. If it is a laser distance sensor, it calculates the distance by emitting a laser beam and measuring the time from the laser being emitted to being reflected back to the sensor. This method has high accuracy and can obtain accurate distance data in a short time. When the laser beam irradiates the target area, part of the laser will be reflected back to the sensor. According to the relationship between the speed of light and the round-trip time (distance = speed of light × round-trip time / 2), the distance between the device and the target can be accurately calculated. Shooting distance information is also indispensable for the normal operation of the intelligent code reading device. At different shooting distances, the size and clarity of the target code in the image will be different. For closer targets, some details of the image may be blurred beyond the resolution range of the sensor; for farther targets, the image may be too small to be recognized. After the shooting distance determination module 8 obtains the distance data, the control module 6 can adjust the focal length, aperture and other parameters of the image acquisition module 3 accordingly. If the distance is close, the aperture can be appropriately reduced to increase the depth of field so that the entire target code can be clearly imaged; if the distance is far, the focal length may need to be increased to magnify the image of the target code. In addition, the shooting distance data will also be fed back to the light source module to adjust the brightness and illumination range of the LED light according to the distance to ensure that the target area has appropriate light intensity. For the image processing module 4, the distance information can be used to correct the image scaling ratio, so that images taken at different distances can be in the best state when decoding and analyzing, thereby improving the code reading ability of the entire device for targets at different distances.

[0042] In some embodiments, the AI-based intelligent code reading device also includes a deep learning model building module 9, which is connected to the control module 6 in communication. The model building module is used to train a deep learning model according to the image samples and the corresponding shooting conditions after collecting image samples under multiple different shooting conditions. The shooting conditions include different ambient brightness, image acquisition angles and distances; the control module 6 is connected to the shooting angle determination module 7 and the shooting distance determination module 8 in communication. The control module 6 is also used to input the acquired shooting angle, shooting distance, ambient light data and the original image data into the model building module to obtain the original image data after image correction under different shooting conditions. Specifically, the core function of the deep learning model building module 9 is to train a model that can adapt to a variety of shooting conditions. In order to achieve this goal, it is first necessary to collect a large number of image samples under different shooting conditions. After collecting a variety of image samples, the training of the deep learning model is started. The model building module will use advanced deep learning algorithms. The deep learning algorithm has the ability to automatically extract image features and is very suitable for processing image data. The image sample is used as input data, and the corresponding shooting conditions (ambient brightness, acquisition angle, distance) are used as marking information. During the training process, the model continuously adjusts the internal weights and parameters to learn the characteristic change patterns of images under different shooting conditions. For example, for changes in ambient brightness, the model will learn the overexposure characteristics of images under strong light and the low contrast characteristics under weak light, and find corresponding processing methods. For changes in the acquisition angle, the model can identify the geometric deformation and perspective changes of the image caused by the angle, and learn how to correct these deformations. When faced with changes in image size and clarity caused by different shooting distances, the model can also gradually master the corresponding compensation strategies. These shooting conditions cover key factors such as different ambient brightness, image acquisition angle and distance. The control module 6 inputs the shooting angle, shooting distance, ambient light data and original image data into the model building module in real time. The model building module corrects the original image data according to the trained deep learning model.

[0043] For example, when the information transmitted by the control module 6 indicates that the current shooting angle is deviated and the environment is dark and the distance is far, the model will perform geometric correction, brightness adjustment and appropriate amplification processing on the image based on the previously learned knowledge. In this way, the corrected original image data can be obtained under different shooting conditions, providing high-quality images for the subsequent image processing module 4 to accurately decode, greatly improving the success rate and accuracy of the equipment in reading codes in complex environments.

[0044] In some embodiments, the image processing module 4 in the AI-based intelligent code reading device includes a decoding model determination unit 10, which includes multiple decoding models, and is used to determine different decoding models for decoding according to the type and characteristics of the identification code. Specifically, first, in order to determine the type and characteristics of the identification code, the decoding model determination unit 10 will perform a preliminary analysis on the input image data. This analysis process involves multiple aspects. On the one hand, start with the overall appearance characteristics of the image, such as the geometric shape of the image. If the image presents a long strip shape with equal width bars and spaces, it is likely to be a barcode; if it is a square or rectangular pattern composed of black and white squares, it may be a QR code. This shape-based judgment is an important basis for preliminary identification of the type of identification code. On the other hand, the texture and pattern details of the image are analyzed. Different types of identification codes have unique texture features, such as the width ratio of bars and spaces in some barcodes has a specific range, and the distribution of black and white squares in the QR code also follows specific coding rules. By extracting and analyzing these texture features, the type of identification code can be further confirmed. In addition, the color information of the image will also be considered. Although most identification codes are black and white, there may be color identification codes in some special application scenarios, and color information can also assist in determining the type. The decoding model determination unit 10 contains multiple decoding models, which are specially constructed for different types of identification codes. For barcodes, there is a one-dimensional barcode decoding model, which is established based on the encoding standard of the barcode. The model includes functional modules such as bar and space width measurement, start and end character identification, and check code verification. For two-dimensional codes, corresponding decoding models are constructed according to different two-dimensional code standards. These models will involve functions such as positioning pattern recognition, version information acquisition, and data encoding format analysis of two-dimensional codes. When constructing these models, a large number of known identification code samples are used for training and optimization to ensure the accuracy and stability of the models. These decoding models are stored in the database or storage structure of the decoding model determination unit 10 so that they can be quickly called when needed. After identifying the type and characteristics of the identification code, the decoding model determination unit 10 will start the model selection mechanism. If it is judged to be a barcode, it will further analyze the characteristics of the barcode, such as whether it is a continuous barcode or a discrete barcode, and select the most matching model from the barcode decoding model set based on these more detailed feature information. The same is true for QR codes. By identifying its version, encoding format and other features, a suitable model is selected from the QR code decoding model for decoding. This method of dynamically selecting a decoding model based on the type and characteristics of the identification code avoids the decoding errors and inefficiency that may be caused by using a single general model. It can make full use of the advantages of different decoding models to accurately and quickly decode various types of identification codes, meet complex and diverse code reading requirements, and enable intelligent code reading equipment to operate efficiently in different application scenarios.

[0045] In some embodiments, the image processing module 4 in the AI-based intelligent code reading device includes an image preprocessing unit 11, an image segmentation unit 12, a feature extraction unit 13 and an identification code decoding unit 14, wherein the image preprocessing unit 11 is used to preprocess the collected original image data to obtain processed image data; the image segmentation unit 12 is used to extract the image area containing the identification code from the processed image data using an image processing algorithm; the feature extraction unit 13 is used to extract the feature information of the identification code from the image area containing the identification code; the identification code decoding unit 14 is used to use the corresponding decoding model based on the feature information to obtain the information contained in the identification code. Specifically, the image preprocessing unit 11 plays a foundational role in the entire code reading process. It mainly realizes the preprocessing of the original image data through the following methods. First, it is grayscale processing. For color images, convert them into grayscale images. This is because color information is not critical for the recognition of identification codes in many cases, and grayscale can reduce the amount of data and simplify subsequent processing steps. Through a specific weighted average algorithm, such as the commonly used one that converts the values ​​of the three RGB channels into a grayscale value based on the sensitivity of the human eye to different colors, the color image is converted into a grayscale image. Secondly, it is noise reduction processing. During the image acquisition process, noise will be generated due to environmental factors (such as interference from electronic equipment, light scattering, etc.). Filtering technology is used to eliminate these noises, such as median filtering. Median filtering can effectively remove salt and pepper noise by replacing the value of each pixel in the image with the median of the pixel values ​​in its neighborhood, making the image smoother and facilitating subsequent processing. There is also contrast enhancement. The contrast of the image is enhanced by adjusting the grayscale histogram of the image. For example, using the histogram equalization algorithm, it redistributes the grayscale values ​​of the image so that the pixel distribution in the entire grayscale range of the image is more uniform, which can highlight the detailed information in the image and make the identification code clearer and more recognizable in the image. After these preprocessing operations, the original image data is converted into processed image data that is more suitable for subsequent processing, providing a good foundation for accurate code reading.

[0046] The main function of the image segmentation unit 12 is to accurately extract the image area containing the identification code from the preprocessed image data. A common method is based on threshold segmentation. By analyzing the grayscale histogram of the image, a suitable threshold is determined. The pixels in the image with a grayscale value greater than the threshold are classified into one category, and those less than the threshold are classified into another category. For the identification code image, it usually has a significant difference in grayscale value from the background, and the identification code area can be distinguished from the background by a suitable threshold. The feature extraction unit 13 focuses on mining the key feature information of the identification code from the segmented image area containing the identification code. For barcodes, features such as the width, positional relationship, and continuity of the bars and spaces are extracted. By scanning the pixels in the image area, the width of the black bars and white spaces is calculated, the proportional relationship between them is determined, and at the same time, whether the stripes are interrupted or abnormal is checked. These features are crucial for identifying the encoding rules of the barcode. For a two-dimensional code, the position, size and direction of its positioning pattern, as well as the distribution characteristics of the module, etc. are extracted. Through a specific template matching algorithm or a method based on geometric features, the three positioning patterns of the two-dimensional code are found, and then the position and direction of the two-dimensional code are determined. Then the distribution pattern of the modules (black and white squares) is analyzed to extract encoding-related features, such as version information, error correction level, etc. These extracted feature information provides the key basis for the final decoding.

[0047] The identification code decoding unit 14 uses the corresponding decoding model to obtain the information contained in the identification code based on the feature information extracted previously. If it is a barcode, the corresponding barcode decoding model is used according to the extracted bar and space features. For common barcode types, such as EAN-13, the model will convert the width of the bar and space into digital codes according to the encoding rules. First, the start and end characters are identified, and then the combination of bars and spaces is parsed into numbers from left to right or in a specific scanning order, and then the correctness of the result is verified by the check code, and finally the information represented by the barcode is obtained. In this way, through the collaborative work of this series of units, the intelligent code reading device can efficiently and accurately obtain the information contained in the identification code from the original image.

[0048] In some embodiments, the AI-based intelligent code reading device also includes a display module 15 for displaying the code reading results and the working status of the device to the user. The display module 15 can adopt advanced display technologies such as liquid crystal display (LCD) or organic light-emitting diode display (OLED). The LCD display controls the transmittance of light by changing the arrangement of liquid crystal molecules under the action of an electric field, thereby displaying an image. When the image processing module 4 completes the decoding of the identification code, the decoding result will be transmitted to the display module 15 through the data transmission line, and the display module 15 presents these results to the user in an intuitive manner. For some special encoding formats, such as QR codes containing link information, the complete URL may be displayed and corresponding prompts may be provided, such as "click to access related web pages". At the same time, if the identification code contains other types of data, such as product numbers, logistics order numbers, etc., they will also be accurately displayed. In this way, the display module 15 enhances the operability and user experience of the device, allowing users to better use and maintain the AI-based intelligent code reading device.

[0049] In some embodiments, the light source module in the AI-based intelligent code reading device includes an LED lamp rotating unit 16 and a reflection compensation unit 17. The LED lamp rotating unit 16 is used to adjust the illumination angle of the LED lamp; the reflection compensation unit 17 is connected to the environmental sensor 5 in communication. The reflection compensation unit 17 is used to monitor the reflection intensity of the target area in real time through the environmental sensor 5. When the reflection intensity is greater than the set threshold, the wavelength, intensity and illumination angle of the LED lamp are adjusted to compensate for the reflection. Specifically, the LED lamp rotating unit is designed to adjust the illumination angle of the LED lamp. Structurally, it may include components such as a small motor, a transmission device, and an angle sensor. The motor is used as a power source and is connected to the mounting structure of the LED lamp through a transmission device (such as a gear, a belt, etc.). The angle sensor is used to monitor the current angle of the LED lamp in real time. When a control instruction is received, the motor starts and drives the transmission device, so that the LED lamp rotates around a specific axis to adjust the angle. In a complex environment, the position and posture of the target area may be different. By rotating the LED lamp, the lighting direction can be optimized to ensure that the light can evenly and effectively cover the target area. For example, when the target code is located on an inclined plane or is partially blocked by other objects, the rotating LED light can change the incident angle of the light, reduce the generation of shadows, and enable the image acquisition module 3 to obtain an original image with uniform illumination.

[0050] In addition, in different installation locations and usage scenarios, the device may need to adjust the lighting angle according to actual conditions. For example, in a narrow space, by adjusting the angle, the light reflection to the surrounding environment can be avoided to cause interference, and the light can be accurately focused on the target code, thereby improving the quality of image acquisition and the success rate of subsequent decoding. The reflective compensation unit 17 works in conjunction with the environmental sensor 5 to achieve real-time monitoring of the reflection intensity of the target area. The environmental sensor 5 can sense the light conditions in the environment, including the intensity of the reflected light. It detects the light intensity distribution of different wavelengths in the reflected light, converts these light signals into electrical signals, and then obtains accurate reflection intensity data through amplification, filtering and other processing. This data will be transmitted to the reflective compensation unit 17 in real time. When the reflection intensity is greater than the set threshold, the reflective compensation unit 17 starts the adjustment mechanism. For the wavelength adjustment of the LED lamp, it is based on the different reflection characteristics of different objects for different wavelengths of light. When the reflection is detected to be too strong, the wavelength of the emitted light is adjusted by changing the driving current of the LED lamp or using a specific optical filter. For example, if some surfaces have strong reflection of blue light, the blue light component can be appropriately reduced and the proportion of other wavelengths of light can be increased. In terms of intensity adjustment, the overall light intensity is reduced by reducing the power of the LED lamp, thereby reducing the intensity of the reflected light. For the adjustment of the illumination angle, similar to the principle of the LED lamp rotation unit, the angle of the LED lamp is changed by adjusting the motor and other driving devices, so that the light enters the target area at a more appropriate angle and avoids direct reflection back to the image acquisition module 3. This comprehensive adjustment method can effectively deal with the reflection problem, so that the image of the target area will not lose details or produce over-bright areas due to reflection during acquisition, ensuring image quality and providing high-quality original image data for subsequent image processing and decoding. In some embodiments, the AI-based intelligent code reading device also includes a power supply module 18 for providing power to various components in the device. Specifically, the power supply module 18 not only provides electrical energy, but also reasonably distributes the power to various components in the device. It realizes accurate distribution of power through internal circuit board wiring and power management chips. Different components may have different requirements for voltage and current. For example, the LED lamp in the light source module may require a higher current to ensure sufficient brightness, while the chip in the image processing module 4 may have higher requirements for voltage stability. The power supply module 18 will adjust the output voltage and current through the power management chip according to the specific needs of each component to ensure that each component can obtain a suitable power supply. At the same time, in order to cope with the voltage fluctuations that may occur during the operation of the device, the power supply module 18 also takes a series of voltage stabilization measures. In addition to the filtering and voltage stabilization circuits in the power adapter and battery discharge circuit mentioned above, a special voltage stabilization chip may also be used. These voltage-stabilizing chips can monitor the output voltage in real time. When the voltage fluctuates, they can quickly make adjustments to stabilize the output voltage within the set range, ensuring that all components can always operate under stable voltage conditions and avoiding performance degradation, abnormal operation, or even damage to components due to unstable voltage.

[0051] In some embodiments, the AI-based intelligent code reading device also includes a network module 19, which is used to realize the connection between each module in the device and the network. The network module 19 is connected to the control module 6. After the code reading is completed, the network module 19 uploads the code reading result to the set cloud. Specifically, the network module 19 is closely connected with the control module 6 to realize data interaction. During the operation of the device, the control module 6 is responsible for coordinating the work of each module and passing the code reading result to the network module 19 after the code reading is completed. This data transmission is realized through the communication bus or specific data interface inside the device. When the network module 19 obtains the code reading result, it starts to upload the data to the set cloud. First, it needs to establish a communication connection with the cloud server. After the connection is established, the network module 19 will encapsulate and encode the code reading result data, convert it into a format suitable for transmission in the network, and then send the data to the cloud server through the network connection. The cloud server can store, analyze and further process the code reading results, such as integrating data with other systems, performing big data analysis, etc., to provide users with richer functions and services, such as logistics information tracking, production data statistics, etc., which greatly improves the value of the device in the entire business process.

[0052] In some embodiments, the AI-based intelligent code reading device also includes an alarm module 20, which is connected to the control module 6 for sending a sound or light effect alarm after receiving the code reading failure information sent by the control module 6. Specifically, a stable communication connection is established between the alarm module 20 and the control module 6. After detecting the code reading failure, the control module 6 will encode the code reading failure information into a data signal in the format specified by the preset protocol, and transmit it to the alarm module 20 through the bus. The communication interface circuit in the alarm module 20 will monitor the bus in real time. Once the information that conforms to the protocol is received, it will be decoded and identified as code reading failure information, thereby triggering an alarm action. A special logic judgment unit is provided inside the alarm module 20 to determine whether the alarm triggering condition is met. After receiving the information from the control module 6, the logic judgment unit will parse and verify the information. It will check whether the source of the information is a legitimate control module 6, and whether the content of the information clearly indicates that the code reading failed. Only when these conditions are met will the alarm mechanism be activated. This logic judgment can effectively prevent the occurrence of false alarms and ensure the accuracy and reliability of the alarm.

[0053] Among them, the alarm module 20 is equipped with a sound-generating device, such as a speaker or a buzzer. When the alarm is triggered, the audio driving circuit in the alarm module 20 sends an electrical signal of a specific frequency and intensity to the sound-generating device. Different sound modes can be set for different types of code reading failures. At the same time, the alarm module 20 also includes light-emitting elements, such as LED lights. These LED lights can indicate code reading failures through different colors and flashing modes. Through this perfect alarm module 20 design, the AI-based intelligent code reading device can quickly notify operators when code reading failures occur, so that they can take timely measures to deal with them, reduce production delays, data errors and other problems that may be caused by code reading failures, and improve the efficiency and reliability of the entire workflow.

[0054] The AI-based intelligent code reading device of this embodiment includes: a shell 1, a light source module 2, an image acquisition module 3, an image processing module 4, an environmental sensor 5, a control module 6, a shooting angle determination module 7, a shooting distance determination module 8, a deep learning model establishment module 9, a decoding model determination unit 10, an image preprocessing unit 11, an image segmentation unit 12, a feature extraction unit 13, an identification code decoding unit 14, a display module 15, an LED light rotation unit 16, a reflection compensation unit 17, a power supply module 18, a network module 19, and an alarm module 20.

[0055] The housing 1 is made of aluminum alloy and is used to protect the internal components; The light source module is set at the front end of the device and has multiple LED lights with different wavelengths built in. The light source module is used to adjust the lighting mode according to actual needs; Image acquisition module 3, using a high-definition CMOS sensor, is used to collect raw image data of the target area; An image processing module 4 is connected to the image acquisition module 3 in communication. The image processing module 4 integrates a high-performance DSP chip. The image processing module 4 is used to analyze the image data in combination with an AI recognition algorithm to obtain a decoding result after receiving the original image data transmitted by the image acquisition module 3; An environmental sensor 5, arranged outside the housing 1, for determining ambient light data; The control module 6 is in communication with the light source module, the image processing module 4 and the environment sensor 5. When the control module 6 receives the device start-up instruction, it determines the ambient light data through the environment sensor 5, and then controls the brightness of the LED light according to the ambient light data. After a set time, it controls the image acquisition module 3 to shoot the target area at a set frequency to obtain original image data, and sends the original image data to the image processing module 4 to obtain a decoding result. A shooting angle determination module 7, which is arranged in the image acquisition module 3 and is used to detect the shooting angle of the image acquisition module 3 in real time; A shooting distance determination module 8, arranged at the front end of the device, is used to determine the shooting distance between the device and the target area; A deep learning model building module 9 is connected to the control module 6 for training a deep learning model according to the image samples and the corresponding shooting conditions after collecting image samples under a plurality of different shooting conditions, wherein the shooting conditions include different ambient brightness, image acquisition angles and distances; The control module 6 is in communication with the shooting angle determination module 7 and the shooting distance determination module 8, and the control module 6 is also used to input the acquired shooting angle, shooting distance, ambient light data and the original image data into the model building module to obtain the original image data after image correction under different shooting conditions; The image processing module 4 includes a decoding model determination unit 10, which includes a plurality of decoding models and is used to determine different decoding models for decoding according to the type and characteristics of the identification code; The image processing module 4 includes an image preprocessing unit 11, an image segmentation unit 12, a feature extraction unit 13 and an identification code decoding unit 14, wherein the image preprocessing unit 11 is used to preprocess the collected original image data to obtain processed image data; The image segmentation unit 12 is used to extract the image area containing the identification code from the processed image data using an image processing algorithm; The feature extraction unit 13 is used to extract feature information of the identification code from the image area containing the identification code; The identification code decoding unit 14 is used to use a corresponding decoding model based on the feature information to obtain information included in the identification code; Display module 15, used to display the code reading result and the working status of the device to the user; The light source module includes an LED lamp rotating unit 16 and a reflection compensation unit 17. The LED lamp rotating unit 16 is used to adjust the irradiation angle of the LED lamp. The reflection compensation unit 17 is connected to the environment sensor 5 for communication. The reflection compensation unit 17 is used to monitor the reflection intensity of the target area in real time through the environment sensor 5. When the reflection intensity is greater than a set threshold, the wavelength, intensity and irradiation angle of the LED light are adjusted to compensate for the reflection. A power supply module 18, used to provide power to various components in the device; The network module 19 is used to realize the connection between each module in the device and the network. The network module 19 is connected to the control module 6. After the code reading is completed, the network module 19 uploads the code reading result to the set cloud; The alarm module 20 is connected to the control module 6 for sending out a sound or light alarm after receiving the code reading failure information from the control module 6 .

[0056] As mentioned above, it is possible to automatically adapt to different lighting environments and dynamically adjust LED lighting parameters to ensure stable and clear images under various lighting conditions. The AI ​​algorithm is then used to accurately identify and decode identification code information, effectively solving the problem of traditional code reading equipment being prone to misidentification in complex environments such as strong light and weak light.

Claims

1. An AI-based intelligent code reading device, characterized in that: include: The housing is made of aluminum alloy to protect the internal components; The light source module is arranged at the front end of the device and has multiple LED lights of different wavelengths built in. The light source module is used to adjust the lighting mode according to actual needs; The image acquisition module uses a high-definition CMOS sensor to collect raw image data of the target area; An image processing module is communicatively connected with the image acquisition module, the image processing module integrates a high-performance DSP chip, and is used for analyzing the image data in combination with an AI recognition algorithm to obtain a decoding result after receiving the original image data transmitted by the image acquisition module; an environmental sensor, disposed outside the housing, for determining ambient light data; The control module is communicatively connected with the light source module, the image processing module and the environmental sensor. When the control module receives the device start-up instruction, it determines the ambient light data through the environmental sensor, and then controls the brightness of the LED light according to the ambient light data. After a set time, it controls the image acquisition module to shoot the target area at a set frequency to obtain raw image data, and sends the raw image data to the image processing module to obtain a decoding result.

2. The AI-based intelligent code reading device according to claim 1, characterized in that: It also includes a shooting angle determination module, which is arranged in the image acquisition module and is used to detect the shooting angle of the image acquisition module in real time; The shooting distance determination module is arranged at the front end of the device and is used to determine the shooting distance between the device and the target area.

3. The AI-based intelligent code reading device according to claim 2, characterized in that: It also includes a deep learning model building module, which is in communication with the control module, and the model building module is used to train a deep learning model according to the image samples and the corresponding shooting conditions after collecting image samples under multiple different shooting conditions, wherein the shooting conditions include different ambient brightness, image acquisition angles and distances; The control module is communicatively connected with the shooting angle determination module and the shooting distance determination module, and the control module is also used to input the acquired shooting angle, shooting distance, ambient light data and original image data into the model building module to obtain the original image data after image correction under different shooting conditions.

4. The AI-based intelligent code reading device according to claim 1, characterized in that: The image processing module includes a decoding model determination unit, which includes a plurality of decoding models and is used to determine different decoding models for decoding according to the type and characteristics of the identification code.

5. The AI-based intelligent code reading device according to claim 1, characterized in that: The image processing module includes an image preprocessing unit, an image segmentation unit, a feature extraction unit and an identification code decoding unit, wherein the image preprocessing unit is used to preprocess the collected original image data to obtain processed image data; The image segmentation unit is used to extract the image area containing the identification code from the processed image data using an image processing algorithm; The feature extraction unit is used to extract feature information of the identification code from the image area containing the identification code; The identification code decoding unit is used to use a corresponding decoding model based on the feature information to obtain information included in the identification code.

6. The AI-based intelligent code reading device according to claim 1, characterized in that: It also includes a display module for displaying the code reading results and the working status of the device to the user.

7. The AI-based intelligent code reading device according to claim 1, characterized in that: The light source module includes an LED lamp rotating unit and a reflection compensation unit, and the LED lamp rotating unit is used to adjust the irradiation angle of the LED lamp; The reflection compensation unit is communicatively connected to the environmental sensor, and is used to monitor the reflection intensity of the target area in real time through the environmental sensor. When the reflection intensity is greater than a set threshold, the wavelength, intensity and irradiation angle of the LED light are adjusted to compensate for the reflection.

8. The AI-based intelligent code reading device according to claim 1, characterized in that: It also includes a power supply module for providing power to various components in the device.

9. The AI-based intelligent code reading device according to claim 1, characterized in that: It also includes a network module for connecting each module in the device with the network. The network module is connected to the control module. After the code reading is completed, the network module uploads the code reading result to the set cloud.

10. The AI-based intelligent code reading device according to claim 1, characterized in that: It also includes an alarm module, which is in communication with the control module and is used to issue a sound or light alarm after receiving the code reading failure information sent by the control module.

Images