Drawing recognition method and system based on artificial intelligence
Through the artificial intelligence-based drawing recognition method, including image preprocessing, feature extraction, text recognition and line detection, the problem of low accuracy of traditional methods on complex drawings is solved, and efficient and accurate drawing information recognition is achieved.
Patent Information
- Application Number
- CN202510704245.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-29
- Publication Date
- 2025-09-16
AI Technical Summary
Existing drawing recognition methods based on deep learning algorithms have low accuracy for poor quality and complex drawings, and are unable to meet the efficiency and accuracy requirements of design drawing review.
An artificial intelligence-based drawing recognition method is adopted, including image preprocessing, feature extraction, image feature judgment, text recognition, line detection and size information determination, and recognition accuracy is improved through convolutional neural networks and deep learning algorithms.
It significantly improves the recognition accuracy and efficiency on complex and poor-quality drawings, can accurately identify drawing information, adapt to drawing images of different complexities, reduce manpower burden, and improve drawing review efficiency.
Smart Images
Figure CN120656199A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of artificial intelligence image recognition, and in particular to an artificial intelligence-based drawing recognition method and system. Background Art
[0002] Currently, most design drawing review relies on manual review, a cumbersome process with high repetitiveness, which inevitably leads to some simple errors being overlooked. To improve the accuracy and efficiency of design drawing review, more and more people are considering applying artificial intelligence technology to recognize drawings before reviewing them.
[0003] Drawing recognition methods and systems based on artificial intelligence algorithms can significantly reduce the time and effort required for drawing review, alleviating the labor burden, freeing up people's time and energy, and improving the accuracy of drawing review. However, existing technologies still have the following problems and shortcomings: Traditional recognition methods based on deep learning algorithms cannot achieve high accuracy for poor-quality and complex drawings. Summary of the Invention
[0004] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a drawing recognition method and system based on artificial intelligence, which is more accurate and effective than traditional drawing recognition methods.
[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows: A drawing recognition method based on artificial intelligence, the key of which is to include the following steps: Step 1: Obtain an image set of drawings to be identified and perform preprocessing; Step 2: Extract features from the preprocessed drawing image to obtain drawing image features, and determine whether it is a recognized drawing based on the drawing image features. If so, retrieve and output the drawing information that matches the drawing image features; otherwise, proceed to step 3. Step 3: Obtain the drawing recognition entropy of the recognized drawing, perform text recognition on the drawing image features, and obtain the character string information of the drawing image; Step 4: Determine the minimum bounding rectangle of each character based on the character string information, and mask the location of the minimum bounding rectangle in the pre-processed drawing image; Step 5: Perform line detection on the masked image; Step 6: Based on the string information and line detection results, determine the size information of each structure in the drawing; Step 7: Convert the character string information, line detection results, and size information into drawing information and output it.
[0006] Furthermore, the steps of pre-processing the image set of drawings to be identified in step 1 are as follows: Performing binarization processing on the image of each drawing to be identified; For the tilted binary images in the binary image set, bilinear interpolation method is used to perform rotation correction; The rotation-corrected binary image set is cleaned and filtered.
[0007] Furthermore, in step 2, feature extraction is performed on the pre-processed drawing image to obtain the drawing image features, specifically including: Performing color histogram conversion on the preprocessed drawing image to obtain color features of the preprocessed drawing image; After normalizing the preprocessed drawing image, a feature vector of the preprocessed drawing image is extracted through a convolutional neural network; The color features and feature vectors of the preprocessed drawing image are combined to obtain drawing image features.
[0008] Furthermore, the process of judging the identified drawings in step 2 is as follows: Obtaining a pending identification character corresponding to the drawing image feature based on the extracted drawing image feature; Obtain all identification characters stored in the drawing information library, and measure the similarity between the pending identification characters and the identification characters in the drawing information library; When the similarity between the identification character existing in the drawing information library and the to-be-determined identification character is higher than a preset similarity threshold, the drawing corresponding to the current drawing image feature is determined to be the recognized drawing.
[0009] Furthermore, the steps for determining the drawing recognition entropy in step 3 are as follows: Obtaining drawing image features of the identified drawing and corresponding drawing information; Extracting identification vectors from the drawing image features and the drawing information to obtain a drawing identification enthalpy vector; Extracting an identification mark sequence from the drawing information to obtain an identification mark correction sequence; The drawing recognition entropy of the recognized drawing is determined according to the recognition mark correction sequence and the drawing recognition enthalpy vector.
[0010] Furthermore, the specific process of identifying the enthalpy vector from the drawing is as follows: aligning the drawing image features of each identified drawing according to the drawing information to obtain a correction feature set; Determine the identification mark enthalpy base corresponding to each calibration feature of the calibration feature set according to the drawing information; The drawing identification enthalpy vector is determined based on the identification mark enthalpy base corresponding to each calibration feature.
[0011] Furthermore, in step 6, determining the size information of each structure in the drawing based on the string information and line detection results specifically includes: Obtain a candidate size number set according to the string information; Parse, locate, and extract the line detection results, remove impure line segments, analyze and process different types of straight lines and their combinations, and output the starting and ending point coordinate information; According to the candidate size number set and the start and end point coordinate information of the different types of straight lines, the size information of each structure in the drawing is determined by utilizing the logical connection between the graphic element and the annotation information.
[0012] In a second aspect, the present invention provides an artificial intelligence-based drawing recognition system, which uses the method described in the first aspect to perform drawing recognition, including: An acquisition module is used to acquire an image set of drawings to be identified and perform preprocessing; a processing module, performing feature extraction on the image set of the drawing to be identified to obtain drawing image features, and determining whether the current drawing is an identified drawing based on the drawing image features; A processing module is used to extract features from the pre-processed drawing image to obtain drawing image features, and determine whether the drawing is a recognized drawing based on the drawing image features. When the current drawing is determined to be a recognized drawing, the drawing information matching the drawing image features is retrieved and output; The processing module is further configured to, when determining that the current drawing is an unrecognized drawing, perform text recognition based on the drawing image features to obtain character string information of the drawing image, determine the minimum bounding rectangle of each character based on the character string information, mask the location of the minimum bounding rectangle in the pre-processed drawing image, and perform line detection on the masked image; The processing module is further used to determine the size information of each structure in the drawing based on the string information and the line detection result; The storage module is used to convert character string information, line detection results, and size information into drawing information and output it.
[0013] In a third aspect, the present invention provides a computer terminal device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to obtain the computer program and execute an artificial intelligence-based drawing recognition method as described in the first aspect.
[0014] In a fourth aspect, the present invention provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to implement the steps of an artificial intelligence-based drawing recognition method as described in the first aspect.
[0015] The remarkable effects of the present invention are: The present invention provides an artificial intelligence-based drawing recognition method and system, which first obtains an image set of drawings to be recognized and performs preprocessing; performs feature extraction on the preprocessed drawing images to obtain drawing image features, and determines whether it is a recognized drawing based on the drawing image features. If so, draw information matching the drawing image features is retrieved and output; if it is an unrecognized drawing, obtains the drawing recognition entropy of the recognized drawing, performs text recognition on the drawing image features, and obtains character string information of the drawing image; determines the minimum bounding rectangle of each character based on the character string information, and masks the position of the minimum bounding rectangle in the preprocessed drawing image; performs line detection on the masked image; determines the size information of each structure in the drawing based on the character string information and the line detection results; forms drawing information with the character string information, line detection results, and size information, and stores the information according to preset storage rules.
[0016] This application first performs data preprocessing or filtering on the image set of drawings to be identified, significantly improving the accuracy and recognition efficiency of the subsequent intelligent recognition algorithm; then, by determining whether the drawings to be identified have been identified, and obtaining the drawing recognition entropy of the identified drawings when it is determined that the drawings have not been identified, the text recognition of the drawing image features is performed, so that the key information in the drawings is more prominent in the drawing images, and the drawing information can still be accurately identified in more complex, more blurred or poorer quality conditions; then, after processing the drawing images using the identified character string information, line edge detection is performed; finally, the dimensional information of each structure is obtained based on the line detection results, thereby forming the drawing information. This effectively improves the recognition algorithm's ability to recognize drawing information, helps to accurately recognize drawing information in more complex, more blurred or poorer quality conditions, effectively improves the algorithm's ability to adapt to drawing images of different complexities, and greatly improves the recognition speed and accuracy of drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 is an exemplary flow chart of the method of the present invention; Figure 2 It is a structural diagram of the system of the present invention; Figure 3 It is a structural schematic diagram of the computer terminal of the present invention. DETAILED DESCRIPTION
[0018] The specific implementation manner and working principle of the present invention will be further described in detail below with reference to the accompanying drawings. Example
[0019] like Figure 1 As shown, the embodiment of the present invention provides a drawing recognition method based on artificial intelligence, and the specific steps are as follows: Step 1: Obtain an image set of drawings to be identified and perform preprocessing; In some specific implementations, the steps of pre-processing the image set of drawings to be identified are as follows: Step 101: binarize the image of each drawing to be identified using the optimal threshold processing based on the Ostu method; Since problems are inevitable during the image acquisition process of the drawing to be identified, it is necessary to find the tilt angle of the binary scanned document image in step 1. In this embodiment, only the tilt angle within the range of plus or minus 5 degrees in the horizontal and vertical directions is detected; Step 102: For the tilted binary images in the binary image set, a bilinear interpolation method is used to perform rotation correction on the tilted images; The specific steps are as follows: In image space, all straight lines passing through a point (x, y) can be expressed using polar coordinate equations: ρ=xcosθ+ysinθ,0≤θ≤π Where ρ is the distance from the origin to the line, and θ is the angle between the line and the x-axis. An accumulator H(ρ,θ) is created in the parameter space ρ,θ. The elements of the accumulator are initialized to 0. For any point (x, y) in the image, all corresponding values of ρ and θ in the parameter space are found. The accumulator H(ρ,θ) is summed up. Local peak detection is then performed on the accumulator H(ρ,θ) to obtain the ρ and θ parameters for all lines in the image. To reduce interference from redundant information, only lines within the image near the upper boundary are considered. The longest line l in this portion of the image is found, and the parameter θ of l is the skew angle of the scanned document. Finally, bilinear interpolation is used to rotate the image to complete the skew correction.
[0020] Step 103: Perform cleaning and filtering processing on the rotation-corrected binary image set.
[0021] Through the above-mentioned processing, the oblique character string is rotated and corrected, thereby solving the problem of character multi-directionality and redundant characters, so that the character area can be accurately located, which helps to accurately extract the character string information in the subsequent steps.
[0022] Step 2: Extract features from the preprocessed drawing image to obtain drawing image features, and determine whether it is a recognized drawing based on the drawing image features. If so, retrieve and output the drawing information that matches the drawing image features; otherwise, proceed to step 3. In some specific implementations, extracting features from the preprocessed drawing image to obtain the drawing image features specifically includes: Step 211: performing color histogram conversion on the preprocessed drawing image to obtain color features of the preprocessed drawing image; In specific implementation, the preprocessed drawing image can be converted from RGB color space to HSV color space, and then the color histogram in HSV color space is calculated, and then the histogram features of different channels are merged to obtain a complete color feature vector as the color feature of the drawing image.
[0023] Step 212: After normalizing the pre-processed drawing image, extracting a feature vector of the pre-processed drawing image through a convolutional neural network; In the specific implementation, we first use a pre-trained convolutional neural network model (such as ResNet50, VGG16, etc.) to extract image features, then use the Keras library to load the pre-trained ResNet50 model, and remove the fully connected layer of the model. Finally, feature extraction is performed on the entire image set, and the feature vectors of all images are stored together to obtain the feature vector of the drawing image.
[0024] Step 213: Combine the color features and feature vectors of the preprocessed drawing image to obtain drawing image features.
[0025] It should be noted that the drawing image feature is the identification feature of the drawing to be identified, which can be used to determine whether the drawing to be identified is a drawing to be identified for the first time, and then to identify the drawing when it is repeatedly acquired, thereby shortening the drawing identification time.
[0026] In this embodiment, the process of determining the recognized drawings is as follows: Step 221: Input the extracted drawing image features into the deep learning algorithm model, and obtain the pending identification characters corresponding to the drawing image features through the following operations. The specific steps are as follows: The convolutional neural network is used to extract the features of the drawing image and convert them into text vector information. The embodiment of the present invention uses a convolutional neural network method, which is more accurate than the traditional template matching method; feature extraction is performed on the text vector information to obtain the specified features of the text vector information; based on the mapping relationship between the features and the identification characters, the undetermined identification characters corresponding to the specified features are obtained; Step 222: Obtain all identification characters stored in the drawing information library, and measure the similarity between the pending identification characters and the identification characters in the drawing information library; Step 223: When the similarity between the identification character in the drawing information database and the identification character to be determined is higher than a preset similarity threshold, the drawing corresponding to the current drawing image feature is determined to be a recognized drawing. In specific implementation, the similarity threshold is calibrated as a constant based on experience.
[0027] Based on the above-mentioned judgment process of the identified drawings, automatic matching can be performed based on the extracted drawing image features and the identification characters in the drawing information library. This not only makes it possible to find the identified drawings conveniently and quickly through the drawing image features, but also ensures the accuracy of the specified drawing information found. It can also avoid the time waste caused by the secondary recognition of the identified drawings, greatly improving the efficiency of drawing recognition.
[0028] Step 3: Obtain the drawing recognition entropy of the recognized drawing, perform text recognition on the drawing image features, and obtain the character string information of the drawing image; In some specific implementations, obtaining drawing image features of the identified drawing and corresponding drawing information; Step 311: extracting identification vectors from the drawing image features and the drawing information to obtain a drawing identification enthalpy vector; Step 312: extract the identification mark sequence from the drawing information to obtain an identification mark correction sequence; Step 313: Determine the drawing recognition entropy of the recognized drawing according to the recognition mark correction sequence and the drawing recognition enthalpy vector.
[0029] Furthermore, in this example, the specific process of obtaining the enthalpy vector by identifying the drawing is as follows: Step 321: align the drawing image features of each identified drawing according to the drawing information to obtain a correction feature set; Step 322: Determine the identification mark enthalpy base corresponding to each calibration feature of the calibration feature set according to the drawing information; Step 323: Determine the drawing identification enthalpy vector according to the identification mark enthalpy basis corresponding to each calibration feature.
[0030] It should be noted that the identification mark enthalpy basis is the relative difference between the image-based drawing recognition result and the drawing information features during the drawing recognition process. Since this relative difference is usually caused by interference from environmental factors during the image acquisition process, the image corresponding to the subsequently acquired drawing can be enhanced according to the identification mark enthalpy basis, thereby reducing the impact of environmental factors on the drawing recognition accuracy during the drawing image acquisition process.
[0031] In specific implementation, the ratio between each image-based drawing recognition result and the drawing information feature can be used as the identification mark enthalpy basis corresponding to the drawing image, and each identification mark enthalpy basis can be used as a vector element to form the drawing recognition enthalpy vector.
[0032] Step 4: Based on the character string information, obtain the corresponding real coordinates on the drawing to be recognized; determine the minimum bounding rectangle of each character according to its real coordinates, and mask the position of the minimum bounding rectangle in the pre-processed drawing image; This step masks the characters and components in the image to reduce the interference of the lines in the characters and components on our recognition results.
[0033] Step 5: Perform line detection on the masked image; After masking the characters and components in the image, use OpenCV's fast line detection to obtain line data. Create a fast line detector class (this detector can be created using the createFastLineDetector function in OpenCV) and use it to obtain line vectors from the image. This section requires traversing the line vectors, obtaining the line coordinates, and saving the x, y values of the line's starting and ending points. When saving, pay attention to the order of the line coordinates: the smaller x value is used as the starting x value, and the larger x value is used as the ending x value. The y value is modified accordingly.
[0034] Step 6: Based on the string information and line detection results, determine the size information of each structure in the drawing; In an embodiment of the present invention, determining the size information of each structure in the drawing specifically includes: Step 601: Obtain a candidate size number set according to the character string information; Step 602: parse, locate, and extract the line detection results, remove impure line segments, analyze and process different types of straight lines and their combinations, and output the start and end point coordinate information; Step 603: Determine the size information of each structure in the drawing based on the candidate size number set and the start and end point coordinate information of the different types of straight lines, using the logical connection between the graphic elements and the annotation information.
[0035] Step 7: The character string information, line detection results, and size information are converted into drawing information and stored and / or output according to a preset storage rule. Example
[0036] like Figure 2 As shown, an embodiment of the present invention provides an artificial intelligence-based drawing recognition system 200, which uses the method described in Example 1 to perform drawing recognition, including: An acquisition module 201 is used to acquire an image set of drawings to be identified and perform preprocessing; Processing module 202 performs feature extraction on the image set of the drawing to be identified to obtain drawing image features, and determines whether the current drawing is an identified drawing based on the drawing image features; Processing module 202 is used to extract features from the pre-processed drawing image to obtain drawing image features, and determine whether the drawing is a recognized drawing based on the drawing image features. If the current drawing is determined to be a recognized drawing, drawing information matching the drawing image features is retrieved and output; The processing module 202 is further configured to, when determining that the current drawing is an unrecognized drawing, perform text recognition based on the drawing image features to obtain character string information of the drawing image, determine the minimum bounding rectangle of each character based on the character string information, mask the location of the minimum bounding rectangle in the pre-processed drawing image, and perform line detection on the masked image; The processing module 202 is further configured to determine the size information of each structure in the drawing based on the character string information and the line detection result; The storage module 203 is used to convert the character string information, line detection results, and size information into drawing information and store it according to a preset storage rule.
[0037] The above describes in detail an example of an artificial intelligence-based drawing recognition method and system provided in an embodiment of the present application. It can be understood that in order to achieve the above functions, the corresponding device includes a hardware structure and / or software module corresponding to executing each function.
[0038] Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function in the application is executed in hardware or in a computer software-driven hardware manner depends on the specific application and design constraints of the technical solution. Therefore, professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application. Example
[0039] like Figure 3 As shown, an embodiment of the present invention provides a computer terminal device 300, which includes a memory and a processor, the memory stores a computer program, and the processor is configured to obtain the computer program and execute an artificial intelligence-based drawing recognition method as described in Example 1.
[0040] In some embodiments, reference Figure 3 , which is a structural diagram of a computer terminal device 300 that applies an artificial intelligence-based drawing recognition method according to some embodiments of the present application. In the above embodiment, an artificial intelligence-based drawing recognition method can be used by Figure 3The computer terminal device 300 shown in FIG. 1 is implemented as shown in FIG. 1 , and the computer terminal device 300 includes at least one communication bus 301 , a communication interface 302 , a processor 303 and a memory 304 .
[0041] The processor 303 can be a general-purpose central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more processors for controlling the execution of an artificial intelligence-based drawing recognition method in the present application.
[0042] The communication bus 301 may include a path for transmitting information between the aforementioned components.
[0043] Memory 304 may be, but is not limited to, a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disk storage, an optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer. Memory 304 may be independent and connected to processor 303 via communication bus 301. Memory 304 may also be integrated with processor 303.
[0044] The memory 304 is used to store program codes for executing the solution of the present application, and the execution is controlled by the processor 303. The processor 303 is used to execute the program codes stored in the memory 304. The program codes may include one or more software modules.
[0045] The communication interface 302 uses any transceiver or other device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area network (WLAN), etc.
[0046] Optionally, the computer terminal device 300 may further include a power supply 305 for providing power to various devices or circuits in the real-time computer terminal device.
[0047] In a specific implementation, as an example, a computer terminal device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor herein may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).
[0048] The aforementioned computer terminal device can be a general-purpose computer terminal device or a dedicated computer terminal device. In a specific implementation, the computer terminal device can be a desktop computer, a portable computer, a network server, a personal digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. The embodiments of this application do not limit the type of computer terminal device. Example
[0049] An embodiment of the present invention provides a computer-readable storage medium, which stores at least one computer program, and is characterized in that the computer program is loaded and executed by a processor to implement the steps of an artificial intelligence-based drawing recognition method as described in Example 1.
[0050] In summary, the present invention discloses an artificial intelligence-based drawing recognition method and system. First, an image set of drawings to be recognized is obtained and preprocessed. Feature extraction is performed on the preprocessed drawing images to obtain drawing image features. Based on the drawing image features, it is determined whether it is a recognized drawing. If so, drawing information matching the drawing image features is retrieved and output. If it is an unrecognized drawing, the drawing recognition entropy of the recognized drawing is obtained. Text recognition is performed on the drawing image features to obtain string information of the drawing image. Based on the string information, the minimum bounding rectangle of each character is determined, and the position of the minimum bounding rectangle in the preprocessed drawing image is masked. Line detection is performed on the masked image. Based on the string information and line detection results, the size information of each structure in the drawing is determined. The string information, line detection results, and size information are formed into drawing information and stored according to a preset storage rule. The method effectively improves the recognition algorithm's ability to recognize drawing information, helps to accurately recognize drawing information even in more complex, more blurred, or poorer quality situations, effectively improves the algorithm's ability to adapt to drawing images of different complexities, and greatly improves the recognition speed and accuracy of drawings.
[0051] The technical solution provided by the present invention is introduced in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, the present invention can also be improved and modified in a number of ways, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.
Claims
1. A drawing recognition method based on artificial intelligence, characterized in that: The steps include: Step 1: Obtain an image set of drawings to be identified and perform preprocessing; Step 2: Extract features from the preprocessed drawing image to obtain drawing image features, and determine whether it is a recognized drawing based on the drawing image features. If so, retrieve and output the drawing information that matches the drawing image features; otherwise, proceed to step 3. Step 3: Obtain the drawing recognition entropy of the recognized drawing, perform text recognition on the drawing image features, and obtain the character string information of the drawing image; Step 4: Determine the minimum bounding rectangle of each character based on the character string information, and mask the location of the minimum bounding rectangle in the pre-processed drawing image; Step 5: Perform line detection on the masked image; Step 6: Based on the string information and line detection results, determine the size information of each structure in the drawing; Step 7: Convert the character string information, line detection results, and size information into drawing information and output it.
2. The method for drawing recognition based on artificial intelligence according to claim 1, characterized in that: The steps for preprocessing the image set of drawings to be identified in step 1 are as follows: Performing binarization processing on the image of each drawing to be identified; For the tilted binary images in the binary image set, bilinear interpolation method is used to perform rotation correction; The rotation-corrected binary image set is cleaned and filtered.
3. The drawing recognition method based on artificial intelligence according to claim 1, characterized in that: In step 2, feature extraction is performed on the preprocessed drawing image to obtain the drawing image features, specifically including: Performing color histogram conversion on the preprocessed drawing image to obtain color features of the preprocessed drawing image; After normalizing the preprocessed drawing image, a feature vector of the preprocessed drawing image is extracted through a convolutional neural network; The color features and feature vectors of the preprocessed drawing image are combined to obtain drawing image features.
4. The drawing recognition method based on artificial intelligence according to claim 1, characterized in that: The process of judging the identified drawings in step 2 is as follows: Obtaining a pending identification character corresponding to the drawing image feature based on the extracted drawing image feature; Obtain all identification characters stored in the drawing information library, and measure the similarity between the pending identification characters and the identification characters in the drawing information library; When the similarity between the identification character existing in the drawing information library and the to-be-determined identification character is higher than a preset similarity threshold, the drawing corresponding to the current drawing image feature is determined to be the recognized drawing.
5. The drawing recognition method based on artificial intelligence according to claim 1, characterized in that: The steps for determining the drawing recognition entropy in step 3 are as follows: Obtaining drawing image features of the identified drawing and corresponding drawing information; Extracting identification vectors from the drawing image features and the drawing information to obtain a drawing identification enthalpy vector; Extracting an identification mark sequence from the drawing information to obtain an identification mark correction sequence; The drawing recognition entropy of the recognized drawing is determined according to the recognition mark correction sequence and the drawing recognition enthalpy vector.
6. The method for drawing recognition based on artificial intelligence according to claim 5, characterized in that: The specific process of obtaining the drawing identification enthalpy vector is as follows: aligning the drawing image features of each identified drawing according to the drawing information to obtain a correction feature set; Determine the identification mark enthalpy base corresponding to each calibration feature of the calibration feature set according to the drawing information; The drawing identification enthalpy vector is determined based on the identification mark enthalpy base corresponding to each calibration feature.
7. The drawing recognition method based on artificial intelligence according to claim 1, characterized in that: In step 6, based on the string information and line detection results, the size information of each structure in the drawing is determined, including: Obtain a candidate size number set according to the string information; Parse, locate, and extract the line detection results, remove impure line segments, analyze and process different types of straight lines and their combinations, and output the starting and ending point coordinate information; According to the candidate size number set and the start and end point coordinate information of the different types of straight lines, the size information of each structure in the drawing is determined by utilizing the logical connection between the graphic element and the annotation information.
8. An artificial intelligence-based drawing recognition system, which uses the method according to any one of claims 1 to 7 to perform drawing recognition, characterized in that: include: An acquisition module is used to acquire an image set of drawings to be identified and perform preprocessing; a processing module, performing feature extraction on the image set of the drawing to be identified to obtain drawing image features, and determining whether the current drawing is an identified drawing based on the drawing image features; A processing module is used to extract features from the pre-processed drawing image to obtain drawing image features, and determine whether the drawing is a recognized drawing based on the drawing image features. When the current drawing is determined to be a recognized drawing, the drawing information matching the drawing image features is retrieved and output; The processing module is further configured to, when determining that the current drawing is an unrecognized drawing, perform text recognition based on the drawing image features to obtain character string information of the drawing image, determine the minimum bounding rectangle of each character based on the character string information, mask the location of the minimum bounding rectangle in the pre-processed drawing image, and perform line detection on the masked image; The processing module is further used to determine the size information of each structure in the drawing based on the string information and the line detection result; The storage module is used to convert character string information, line detection results, and size information into drawing information and output it.
9. A computer terminal device, characterized in that: The computer terminal device includes a memory and a processor, the memory stores a computer program, and the processor is configured to obtain the computer program and execute the artificial intelligence-based drawing recognition method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing at least one computer program, characterized in that: The computer program is loaded and executed by a processor to implement the steps of the drawing recognition method based on artificial intelligence as described in any one of claims 1 to 7.