Drawing processing method and system based on scene text recognition

Abstract

This invention provides a drawing processing method and system based on scene text recognition. It identifies all graphic and non-graphic elements within the global scope of a drawing. Based on the shape features of the graphic elements, it divides the drawing into several sets of non-graphic elements. Based on the connected components of all non-graphic elements within each set, it generates several text strings corresponding to those sets and performs fuzzy classification to identify the text results of the non-graphic element sets. Based on the spatial distribution information of the graphic and non-graphic elements on the drawing, it preprocesses the text results and overlays them onto the corresponding areas of the drawing, distinguishing between non-graphic and graphic elements. It also performs connectivity recognition on the non-graphic elements, initially labeling the text strings within them to obtain their corresponding text results. This achieves synchronous association recognition of graphics and text within the drawing, improving the efficiency and accuracy of drawing editing.

Description

Technical Field

[0001] This invention relates to the field of image recognition, and more particularly to a drawing processing method and system based on scene text recognition. Background Technology

[0002] Existing architectural or engineering drafting software focuses on extracting shape information from graphic elements on drawings and importing this information into the software interface for editing. However, it can only provide users with the graphic elements on the drawings; it cannot accurately recognize non-graphical character information such as annotations or descriptions, nor can it import this character information one-to-one into the software interface. This results in users not having comprehensive information during the drawing editing process, hindering accurate editing. Furthermore, considering the complexity and minute detail of architectural or engineering drawings, directly loading text recognition tools into the software would require these tools to recognize the entire drawing, increasing the workload and affecting the software's smooth operation. It also cannot guarantee that the text recognition tool correctly recognizes all character information on the drawing, compromising compatibility with subsequent user editing within the software interface. Therefore, synchronous text character recognition of drawings tailored to the specific application scenarios of drafting software is crucial for improving the efficiency and accuracy of drawing editing. Summary of the Invention

[0003] The purpose of this invention is to provide a drawing processing method and system based on scene text recognition. This method identifies all graphic and non-graphic elements within the global scope of a drawing. Based on the shape features of the graphic elements, it divides the drawing into several sets of non-graphic elements. Based on the connected components of all non-graphic elements within each set, it generates several text strings corresponding to those sets and performs fuzzy classification to identify the text results of the non-graphic element sets. Based on the spatial distribution information of the graphic and non-graphic elements on the drawing, it preprocesses the text results and adds them to the corresponding areas of the drawing. This distinguishes between the non-graphic and graphic elements, performs connectivity recognition on the non-graphic elements, and initially identifies the text strings within them to obtain their corresponding text results. This ensures that the text results accurately and completely represent the meaning of the text, achieving synchronous association recognition of graphics and text within the drawing, improving the efficiency and accuracy of drawing editing, and enabling correct recognition of text content across the entire drawing scope.

[0004] This invention is achieved through the following technical solution:

[0005] Drawing processing methods based on scene text recognition include:

[0006] The drawing is identified to obtain all graphic elements and all non-graphic elements within the global scope of the drawing; based on the shape characteristics of the graphic elements, all non-graphic elements are divided into several non-graphic element sets;

[0007] Based on the connected components of all non-graphical elements under the non-graphical element set, all non-graphical elements are grouped to generate several text strings corresponding to the non-graphical element set; the text strings are then subjected to fuzzy classification and recognition to obtain the text result of the non-graphical element set.

[0008] Based on the spatial distribution information of the graphic elements and non-graphic elements on the drawing, the text results are preprocessed in shape and then overlaid onto the corresponding area of ​​the drawing.

[0009] Optionally, the drawing is identified to obtain all graphic elements and all non-graphic elements within the global scope of the drawing; based on the shape characteristics of the graphic elements, all non-graphic elements are divided into several non-graphic element sets, including:

[0010] Contour recognition is performed on the drawing to obtain the contour vector information of each element within the global scope of the drawing; based on the contour vector information, the boundary contour curvature change characteristics of each element are determined; wherein, the boundary contour curvature change characteristics refer to the difference in curvature change of the boundary contour of the element within a unit length;

[0011] Based on the boundary contour curvature variation characteristics, the elements are classified into graphic elements or non-graphic elements;

[0012] Based on the boundary contour position of the graphic element, determine the relative orientation distribution information of each non-graphic element with respect to the boundary contour of the graphic element; based on the relative orientation distribution information, divide all non-graphic elements into several non-graphic element sets.

[0013] Optionally, based on the boundary contour curvature variation characteristics, the elements are classified into graphic elements or non-graphic elements, including:

[0014] Extract the difference in curvature change of the boundary profile per unit length from the boundary profile curvature change characteristics;

[0015] Using 50 pixels as the reference length for the engineering drawing, the curvature change difference of the boundary contour within each unit length is standardized using the reference length to obtain the standardized value corresponding to the curvature change difference of the boundary contour within each unit length.

[0016] The standardized value corresponding to the curvature change difference of the boundary contour within each unit length is obtained by the following formula:

[0017]

[0018] Where ΔC represents the standardized value corresponding to the difference in curvature of the boundary profile per unit length; ΔC x L represents the curvature variation difference of the boundary profile per unit length; L represents the actual total length of the boundary profile; L0 represents the reference length of the engineering drawing; C max This represents the maximum theoretically possible difference in curvature variation in engineering drawings.

[0019] The standard average value of the curvature change difference of the boundary profile is obtained by using the standardized value corresponding to the curvature change difference of the boundary profile within each unit length.

[0020] The standardized values ​​corresponding to the curvature change difference of the boundary contour within each unit length are compared with the preset standardized thresholds. The boundary contour positions with standardized values ​​not lower than the preset standardized thresholds are selected as high-band positions.

[0021] The number of high-band positions and the standardized values ​​corresponding to the curvature change difference of each high-band position are retrieved;

[0022] The high-band influence factor is obtained by using the number of high-band locations and the standardized value corresponding to the curvature change difference of each high-band location.

[0023] The high-band impact factor is obtained using the following formula:

[0024]

[0025] Where R represents the high-band influence factor; n represents the number of high-band locations; P represents the percentage of high-band locations relative to the total number of locations per unit length of the boundary profile; ΔC i ΔC represents the standardized value corresponding to the curvature change difference at the i-th high-band position; p This represents the standard mean of the difference in curvature variation;

[0026] The comprehensive evaluation coefficient of curvature fluctuation is obtained by combining the high-band influence factor with the standard average value of curvature change difference.

[0027] The comprehensive evaluation coefficient of curvature fluctuation is obtained by the following formula:

[0028]

[0029] Where K represents the comprehensive evaluation coefficient of curvature fluctuation; R represents the high-band influence factor; ΔC p This represents the standard mean of the difference in curvature variation;

[0030] The comprehensive evaluation coefficient of curvature fluctuation is compared with a preset coefficient threshold. If the comprehensive evaluation coefficient of curvature fluctuation exceeds the preset coefficient threshold, it is determined to be a non-graphic element; otherwise, it is determined to be a graphic element.

[0031] Optionally, based on the connected components of all non-graphical elements under the non-graphical element set, all non-graphical elements are grouped to generate several text strings corresponding to the non-graphical element set; fuzzy classification and recognition are performed on the text strings to obtain the text result of the non-graphical element set, including:

[0032] Adaptive color reduction processing is performed on all non-graphic elements under the non-graphic element set to extract the connected components of all non-graphic elements; based on the connected components, spatial adjacency recognition is performed on all non-graphic elements to group all non-graphic elements into several text strings corresponding to the non-graphic element set.

[0033] The text string is subjected to fuzzy classification and recognition to obtain the text information contained in the text string, and the text information is subjected to logical correction processing to obtain the text result of the non-graphic element set.

[0034] Optionally, based on the spatial distribution information of the graphic elements and the non-graphic elements on the drawing, the text result is preprocessed in terms of shape and then overlaid onto the corresponding area of ​​the drawing, including:

[0035] The spatial distribution of the graphic elements and the non-graphic elements is identified within the global scope of the drawing to obtain the spatial occupancy information of each graphic element and the non-graphic element.

[0036] Based on the spatial occupancy information of the graphic elements and the non-graphic elements, the element size ratio between the graphic elements and the non-graphic elements is determined, and the font size of the text result is preprocessed accordingly; then, based on the spatial occupancy information of the non-graphic elements, the preprocessed text result is overlaid and added to the corresponding area of ​​the drawing.

[0037] A drawing processing system based on scene text recognition includes:

[0038] The drawing recognition module is used to recognize the drawing and obtain all graphic elements and all non-graphic elements within the global scope of the drawing;

[0039] The element division module is used to divide all non-graphic elements into several non-graphic element sets based on the shape characteristics of the graphic elements.

[0040] The string generation module is used to group all non-graphic elements based on the connected components of all non-graphic elements under the non-graphic element set, and generate several text strings corresponding to the non-graphic element set.

[0041] The text result determination module is used to perform fuzzy classification and recognition on the text string to obtain the text result of the non-graphic element set;

[0042] The text result preprocessing and adding module is used to preprocess the text result in shape and then overlay it onto the corresponding area of ​​the drawing based on the spatial distribution information of the graphic elements and the non-graphic elements on the drawing.

[0043] Optionally, the drawing recognition module is used to recognize the drawing and obtain all graphic elements and all non-graphic elements within the global scope of the drawing, including:

[0044] Contour recognition is performed on the drawing to obtain the contour vector information of each element within the global scope of the drawing; based on the contour vector information, the boundary contour curvature change characteristics of each element are determined; wherein, the boundary contour curvature change characteristics refer to the difference in curvature change of the boundary contour of the element within a unit length;

[0045] Based on the boundary contour curvature variation characteristics, the elements are classified into graphic elements or non-graphic elements;

[0046] The element partitioning module is used to divide all non-graphical elements into several non-graphical element sets based on the shape characteristics of the graphic elements, including:

[0047] Based on the boundary contour position of the graphic element, determine the relative orientation distribution information of each non-graphic element with respect to the boundary contour of the graphic element; based on the relative orientation distribution information, divide all non-graphic elements into several non-graphic element sets.

[0048] Optionally, based on the boundary contour curvature variation characteristics, the elements are classified into graphic elements or non-graphic elements, including:

[0049] Extract the difference in curvature change of the boundary profile per unit length from the boundary profile curvature change characteristics;

[0050] Using 50 pixels as the reference length for the engineering drawing, the curvature change difference of the boundary contour within each unit length is standardized using the reference length to obtain the standardized value corresponding to the curvature change difference of the boundary contour within each unit length.

[0051] The standardized value corresponding to the curvature change difference of the boundary contour within each unit length is obtained by the following formula:

[0052]

[0053] Where ΔC represents the standardized value corresponding to the difference in curvature of the boundary profile per unit length; ΔC x L represents the curvature variation difference of the boundary profile per unit length; L represents the actual total length of the boundary profile; L0 represents the reference length of the engineering drawing; C max This represents the maximum theoretically possible difference in curvature variation in engineering drawings.

[0054] The standard average value of the curvature change difference of the boundary profile is obtained by using the standardized value corresponding to the curvature change difference of the boundary profile within each unit length.

[0055] The standardized values ​​corresponding to the curvature change difference of the boundary contour within each unit length are compared with the preset standardized thresholds. The boundary contour positions with standardized values ​​not lower than the preset standardized thresholds are selected as high-band positions.

[0056] The number of high-band positions and the standardized values ​​corresponding to the curvature change difference of each high-band position are retrieved;

[0057] The high-band influence factor is obtained by using the number of high-band locations and the standardized value corresponding to the curvature change difference of each high-band location.

[0058] The high-band impact factor is obtained using the following formula:

[0059]

[0060] Where R represents the high-band influence factor; n represents the number of high-band locations; P represents the percentage of high-band locations relative to the total number of locations per unit length of the boundary profile; ΔC i ΔC represents the standardized value corresponding to the curvature change difference at the i-th high-band position; p This represents the standard mean of the difference in curvature variation;

[0061] The comprehensive evaluation coefficient of curvature fluctuation is obtained by combining the high-band influence factor with the standard average value of curvature change difference.

[0062] The comprehensive evaluation coefficient of curvature fluctuation is obtained by the following formula:

[0063]

[0064] Where K represents the comprehensive evaluation coefficient of curvature fluctuation; R represents the high-band influence factor; ΔC pThis represents the standard mean of the difference in curvature variation;

[0065] The comprehensive evaluation coefficient of curvature fluctuation is compared with a preset coefficient threshold. If the comprehensive evaluation coefficient of curvature fluctuation exceeds the preset coefficient threshold, it is determined to be a non-graphic element; otherwise, it is determined to be a graphic element.

[0066] Optionally, the string generation module is used to group all non-graphical elements based on the connected components of all non-graphical elements under the non-graphical element set, and generate several text strings corresponding to the non-graphical element set, including:

[0067] Adaptive color reduction processing is performed on all non-graphic elements under the non-graphic element set to extract the connected components of all non-graphic elements; based on the connected components, spatial adjacency recognition is performed on all non-graphic elements to group all non-graphic elements into several text strings corresponding to the non-graphic element set.

[0068] The text result determination module is used to perform fuzzy classification and recognition on the text string to obtain the text result of the non-graphic element set, including:

[0069] The text string is subjected to fuzzy classification and recognition to obtain the text information contained in the text string, and the text information is subjected to logical correction processing to obtain the text result of the non-graphic element set.

[0070] Optionally, the text result preprocessing and adding module is used to preprocess the text result based on the spatial distribution information of the graphic elements and the non-graphic elements on the drawing, and then overlay and add it to the corresponding area of ​​the drawing, including:

[0071] The spatial distribution of the graphic elements and the non-graphic elements is identified within the global scope of the drawing to obtain the spatial occupancy information of each graphic element and the non-graphic element.

[0072] Based on the spatial occupancy information of the graphic elements and the non-graphic elements, the element size ratio between the graphic elements and the non-graphic elements is determined, and the font size of the text result is preprocessed accordingly; then, based on the spatial occupancy information of the non-graphic elements, the preprocessed text result is overlaid and added to the corresponding area of ​​the drawing.

[0073] Compared with the prior art, the present invention has the following beneficial effects:

[0074] This application provides a drawing processing method and system based on scene text recognition. It identifies all graphic and non-graphic elements within the global scope of a drawing. Based on the shape features of the graphic elements, it divides the drawing into several sets of non-graphic elements. Based on the connected components of all non-graphic elements within each set, it generates several text strings corresponding to those sets and performs fuzzy classification to identify the text results of the non-graphic element sets. Based on the spatial distribution information of the graphic and non-graphic elements on the drawing, it preprocesses the text results and adds them to the corresponding areas of the drawing, distinguishing between non-graphic and graphic elements. It also performs connectivity recognition on the non-graphic elements to initially label the text strings within them, thereby obtaining their corresponding text results. This ensures that the text results accurately and completely represent the meaning of the text, achieving synchronous association and recognition of graphics and text within the drawing, improving the efficiency and accuracy of drawing editing and processing, and enabling correct recognition of text content across the entire drawing scope. Attached Figure Description

[0075] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0076] Figure 1 This is a flowchart illustrating the drawing processing method based on scene text recognition provided by the present invention.

[0077] Figure 2 This is a schematic diagram of the drawing processing system based on scene text recognition provided by the present invention. Detailed Implementation

[0078] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, it should be noted that, for ease of description, only the parts relevant to this application are shown in the accompanying drawings, not the entire structure. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this application.

[0079] The terms “comprising” and “having”, and any variations thereof, used in this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0080] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0081] Please see Figure 1 As shown, an embodiment of this application provides a drawing processing method based on scene text recognition. This drawing processing method based on scene text recognition includes:

[0082] The drawing is identified to obtain all graphic elements and all non-graphic elements within the global scope of the drawing; based on the shape characteristics of the graphic elements, all non-graphic elements are divided into several non-graphic element sets.

[0083] Based on the connected components of all non-graphical elements under the non-graphical element set, all non-graphical elements are grouped to generate several text strings corresponding to the non-graphical element set; fuzzy classification and recognition are performed on the text strings to obtain the text result of the non-graphical element set.

[0084] Based on the spatial distribution information of graphic and non-graphic elements on the drawing, the text results are preprocessed in shape and then overlaid onto the corresponding areas of the drawing.

[0085] The beneficial effects of the above embodiments are that the drawing processing method based on scene text recognition identifies all graphic elements and all non-graphic elements within the global scope of the drawing. Based on the shape features of the graphic elements, it divides the drawing into several sets of non-graphic elements. Based on the connected components of all non-graphic elements under each set, it generates several text strings corresponding to the non-graphic element sets and identifies the text results of the non-graphic element sets through fuzzy classification from the text strings. Based on the spatial distribution information of graphic and non-graphic elements on the drawing, it preprocesses the text results and adds them to the corresponding areas of the drawing, distinguishing between non-graphic elements and graphic elements in the drawing, and performing connectivity recognition on the non-graphic elements to initially label the text strings within them, thereby obtaining their corresponding text results. This ensures that the text results accurately and completely represent the meaning of the text, achieving synchronous association recognition of graphics and text within the drawing, improving the efficiency and accuracy of drawing editing and processing, and achieving correct recognition of text content within the global scope of the drawing.

[0086] In another embodiment, the drawing is identified to obtain all graphic elements and all non-graphic elements within the global scope of the drawing; based on the shape characteristics of the graphic elements, all non-graphic elements are divided into several non-graphic element sets, including:

[0087] Contour recognition is performed on the drawing to obtain the contour vector information of each element within the global scope of the drawing; based on the contour vector information, the boundary contour curvature change characteristics of each element are determined; whereby the boundary contour curvature change characteristics refer to the difference in curvature change of the element's boundary contour within a unit length.

[0088] Based on the characteristics of boundary contour curvature variation, elements are classified into graphic elements or non-graphic elements;

[0089] Based on the boundary contour position of the graphic elements, determine the relative orientation distribution information of each non-graphic element to the boundary contour of the graphic elements; based on the relative orientation distribution information, divide all non-graphic elements into several non-graphic element sets.

[0090] The beneficial effects of the above embodiments are that, during the architectural or engineering drawing process, the resulting drawing includes graphic elements and non-graphic elements. Graphic elements can be, but are not limited to, architectural or engineering drawings, while non-graphic elements can be, but are not limited to, text used for annotation and explanation of the graphics. Considering the significant differences in the outline shapes of graphic and non-graphic elements, outline recognition is performed on the entire drawing to distinguish the types of all elements within that area. To improve the accuracy of distinguishing between graphic and non-graphic elements on the drawing, the outline vector information of each element is extracted from the entire drawing. This outline vector information characterizes the spatial distribution direction characteristics of all boundary outlines of each element; it is a commonly used physical quantity for outline characterization in this field and will not be described in detail here. The outline changes of graphics and non-graphic elements (such as text) differ significantly. Generally, the outline changes of graphic elements are relatively gradual, while the outline changes of non-graphic elements are more rapid. Analyzing the outline vector information determines the curvature change characteristics of the boundary outline of each element, thereby quantifying the difference in curvature change per unit length of the boundary outline of each element. If the curvature change difference of an element's boundary contour within a unit length is less than a preset difference threshold, the element is determined to be a graphic element; otherwise, the element is determined to be a non-graphic element, facilitating subsequent text character-level recognition only for non-graphic elements. Furthermore, non-graphic elements in drawings are used to identify and interpret graphic elements; that is, there is a correlation between non-graphic elements and graphic elements. Correspondingly, the positional relationship between non-graphic elements and their associated graphic elements on the drawing is relatively close, and all non-graphic elements associated with the same graphic element are usually concentrated within a certain directional range corresponding to the same graphic element. To centrally organize and classify all non-graphic elements associated with the same graphic element, based on the boundary contour position of the graphic element, the relative directional distribution information of each non-graphic element to the boundary contour of the graphic element is determined. This divides all non-graphic elements into several non-graphic element sets, ensuring that all non-graphic elements within the same set are concentrated within the aforementioned directional range corresponding to the graphic element, while also guaranteeing the correlation between all non-graphic elements within the same set.

[0091] In another embodiment, based on the boundary contour curvature variation characteristics, the elements are distinguished into graphic elements or non-graphic elements, including:

[0092] Extract the difference in curvature change of the boundary profile per unit length from the boundary profile curvature change characteristics;

[0093] Using 50 pixels as the reference length for the engineering drawing, the curvature change difference of the boundary contour within each unit length is standardized using the reference length to obtain the standardized value corresponding to the curvature change difference of the boundary contour within each unit length.

[0094] The standardized value corresponding to the curvature change difference of the boundary contour within each unit length is obtained by the following formula:

[0095]

[0096] Where ΔC represents the standardized value corresponding to the difference in curvature of the boundary profile per unit length; ΔC x L represents the curvature variation difference of the boundary profile per unit length; L represents the actual total length of the boundary profile; L0 represents the reference length of the engineering drawing; C max This represents the maximum theoretically possible difference in curvature variation in engineering drawings.

[0097] The standard average value of the curvature change difference of the boundary profile is obtained by using the standardized value corresponding to the curvature change difference of the boundary profile within each unit length.

[0098] The standardized values ​​corresponding to the curvature change difference of the boundary contour within each unit length are compared with the preset standardized thresholds. The boundary contour positions with standardized values ​​not lower than the preset standardized thresholds are selected as high-band positions.

[0099] The number of high-band positions and the standardized values ​​corresponding to the curvature change difference of each high-band position are retrieved;

[0100] The high-band influence factor is obtained by using the number of high-band locations and the standardized value corresponding to the curvature change difference of each high-band location.

[0101] The high-band impact factor is obtained using the following formula:

[0102]

[0103] Where R represents the high-band influence factor; n represents the number of high-band locations; P represents the percentage of high-band locations relative to the total number of locations per unit length of the boundary profile; ΔC i ΔC represents the standardized value corresponding to the curvature change difference at the i-th high-band position; p This represents the standard mean of the difference in curvature variation;

[0104] The comprehensive evaluation coefficient of curvature fluctuation is obtained by combining the high-band influence factor with the standard average value of curvature change difference.

[0105] The comprehensive evaluation coefficient of curvature fluctuation is obtained by the following formula:

[0106]

[0107] Where K represents the comprehensive evaluation coefficient of curvature fluctuation; R represents the high-band influence factor; ΔC p This represents the standard mean of the difference in curvature variation;

[0108] The comprehensive evaluation coefficient of curvature fluctuation is compared with a preset coefficient threshold. If the comprehensive evaluation coefficient of curvature fluctuation exceeds the preset coefficient threshold, it is determined to be a non-graphic element; otherwise, it is determined to be a graphic element.

[0109] The beneficial effects of the above embodiments are achieved by eliminating interference from elements of different sizes and extreme curvature values ​​through the coordinated standardization of the difference between the baseline length, actual total length, and theoretical maximum curvature change. This unifies the evaluation criteria, enabling graphic and non-graphic elements of different specifications to have comparable curvature feature representations, thus improving the fairness and accuracy of classification. The standard average value of the curvature change difference comprehensively reflects the overall level of curvature change in the element boundary contour. The high-band influence factor accurately characterizes the local concentration characteristics of curvature fluctuations by integrating the proportion of high-band elements and the deviation of the standardized value of a single high-band element relative to the average value. The dual-dimensional features synergistically cover both overall and local curvature features, overcoming the limitations of single feature representation. The comprehensive evaluation coefficient of curvature fluctuation integrates the high-band influence factor and the standard average value of the curvature change difference, amplifying the feature differences between graphic elements (gentle curvature, few high fluctuations) and non-graphic elements (drastic curvature, concentrated high fluctuations), significantly improving the distinguishability between the two types of elements and reducing the risk of misjudging elements with ambiguous boundaries. Simultaneously, the above technical solutions effectively improve the stability and reliability of classification results. Furthermore, the technical solution is adapted to the element characteristics of architectural and engineering drawings, specifically capturing the high curvature fluctuation characteristics of non-graphical elements such as text annotations and the gentle curvature characteristics of graphic elements. The classification logic closely matches the actual element attributes of the drawings, significantly improving the adaptability and accuracy of element classification in engineering drawing scenarios. All calculation processes are based on basic arithmetic operations and statistical analysis, without complex algorithms or iterative processes. In engineering applications, it can be quickly implemented using conventional image processing modules, balancing classification accuracy and implementation efficiency, and possessing strong engineering practicality and scalability.

[0110] In another embodiment, based on the connected components of all non-graphical elements under the non-graphical element set, all non-graphical elements are grouped to generate several text strings corresponding to the non-graphical element set; fuzzy classification and recognition are performed on the text strings to obtain the text result of the non-graphical element set, including:

[0111] Adaptive color reduction processing is performed on all non-graphic elements under the non-graphic element set to extract the connected components of all non-graphic elements; based on the connected components, spatial adjacency recognition is performed on all non-graphic elements to group all non-graphic elements into several text strings corresponding to the non-graphic element set.

[0112] The text string is fuzzy classified and identified to obtain the text information contained in the text string. The text information is then logically corrected to obtain the text result of the non-graphic element set.

[0113] The beneficial effects of the above embodiments are that the correlation between all non-graphical elements under the non-graphical element set is not the same. There are corresponding connected components between any two non-graphical elements, and the connectivity of these components is related to the text strings to which the two non-graphical elements belong. Generally speaking, the connected components of two non-graphical elements belonging to the same text string are significantly different from those of two non-graphical elements belonging to different text strings. Therefore, adaptive color reduction processing is performed on all non-graphical elements under the non-graphical element set to extract the connected components between all non-graphical elements. These connected components are then analyzed to determine the spatial adjacency relationship (e.g., spatial spacing) between any two non-graphical elements. This grouping of all non-graphical elements into several text strings corresponding to the non-graphical element set ensures that each text string completely contains the corresponding text characters. Furthermore, fuzzy classification and recognition are performed on the text strings to obtain the text information contained within them. Semantic and word order logic corrections are then applied to this text information to obtain the final text result of the non-graphical element set, ensuring the accuracy of the text result.

[0114] In another embodiment, based on the spatial distribution information of graphic and non-graphic elements on the drawing, the text result is preprocessed in terms of form and then overlaid onto the corresponding area of ​​the drawing, including:

[0115] The spatial distribution of graphic elements and non-graphic elements is identified within the global scope of the drawing to obtain the spatial occupancy information of each graphic element and non-graphic element.

[0116] Based on the spatial occupancy information of graphic and non-graphic elements, the element size ratio between graphic and non-graphic elements is determined, and the font size of the text result is preprocessed accordingly. Then, based on the spatial occupancy information of non-graphic elements, the preprocessed text result is added to the corresponding area of ​​the drawing.

[0117] The beneficial effects of the above embodiments are that all graphic elements and all non-graphic elements in a drawing occupy a certain proportion of space within the global area of ​​the drawing. If the corresponding area occupied by the text result after being re-added to the drawing differs significantly from the original area occupied by the non-graphic elements, or if the proportion of the corresponding area occupied by the text result after being re-added to the drawing differs significantly from the original proportion, the visual effect of the text result in the drawing will be reduced, resulting in visual disharmony between the original graphic elements and the text result. Therefore, spatial distribution identification of graphic elements and non-graphic elements within the global area of ​​the drawing is performed to obtain the spatial occupancy information of each graphic element and non-graphic element. Based on the spatial occupancy information of graphic elements and non-graphic elements, the element size ratio between graphic elements and non-graphic elements is determined. This is used to preprocess the text result by adjusting its font size. Based on the spatial occupancy information of non-graphic elements, the preprocessed text result is then overlaid onto the corresponding area of ​​the drawing, thereby maintaining visual harmony between the text result overlaid on the drawing and the original graphic elements on the drawing.

[0118] Please see Figure 2 As shown, an embodiment of this application provides a drawing processing system based on scene text recognition. This drawing processing system based on scene text recognition includes:

[0119] The drawing recognition module is used to recognize drawings and obtain all graphic elements and all non-graphic elements within the global scope of the drawing;

[0120] The element partitioning module is used to divide all non-graphical elements into several non-graphical element sets based on the shape characteristics of graphic elements.

[0121] The string generation module is used to group all non-graphical elements based on the connected components of all non-graphical elements under the non-graphical element set, and generate several text strings corresponding to the non-graphical element set.

[0122] The text result determination module is used to perform fuzzy classification and recognition on text strings to obtain text results for non-graphic element sets;

[0123] The text result preprocessing and adding module is used to preprocess text results based on the spatial distribution information of graphic and non-graphic elements on the drawing, and then overlay and add them to the corresponding areas of the drawing.

[0124] The beneficial effects of the above embodiments are that the scene-based text recognition-based drawing processing system identifies all graphic elements and all non-graphic elements within the global scope of the drawing. Based on the shape features of the graphic elements, it divides the drawing into several sets of non-graphic elements. Based on the connected components of all non-graphic elements under each set, it generates several text strings corresponding to the non-graphic element sets and identifies the text results of the non-graphic element sets through fuzzy classification from the text strings. Based on the spatial distribution information of graphic and non-graphic elements on the drawing, it preprocesses the text results and adds them to the corresponding areas of the drawing, distinguishing between non-graphic elements and graphic elements, and performing connectivity recognition on the non-graphic elements to initially label the text strings within them, thereby obtaining their corresponding text results. This ensures that the text results accurately and completely represent the meaning of the text, achieving synchronous association recognition of graphics and text within the drawing, improving the efficiency and accuracy of drawing editing and processing, and realizing correct recognition of text content within the global scope of the drawing.

[0125] In another embodiment, the drawing recognition module is used to recognize the drawing and obtain all graphic elements and all non-graphic elements within the global scope of the drawing, including:

[0126] Contour recognition is performed on the drawing to obtain the contour vector information of each element within the global scope of the drawing; based on the contour vector information, the boundary contour curvature change characteristics of each element are determined; whereby the boundary contour curvature change characteristics refer to the difference in curvature change of the element's boundary contour within a unit length.

[0127] Based on the characteristics of boundary contour curvature variation, elements are classified into graphic elements or non-graphic elements;

[0128] The element partitioning module is used to divide all non-graphical elements into several non-graphical element sets based on the shape characteristics of graphic elements, including:

[0129] Based on the boundary contour position of the graphic elements, determine the relative orientation distribution information of each non-graphic element to the boundary contour of the graphic elements; based on the relative orientation distribution information, divide all non-graphic elements into several non-graphic element sets.

[0130] The beneficial effects of the above embodiments are that, during the architectural or engineering drawing process, the resulting drawing includes graphic elements and non-graphic elements. Graphic elements can be, but are not limited to, architectural or engineering drawings, while non-graphic elements can be, but are not limited to, text used for annotation and explanation of the graphics. Considering the significant differences in the outline shapes of graphic and non-graphic elements, outline recognition is performed on the entire drawing to distinguish the types of all elements within that area. To improve the accuracy of distinguishing between graphic and non-graphic elements on the drawing, the outline vector information of each element is extracted from the entire drawing. This outline vector information characterizes the spatial distribution direction characteristics of all boundary outlines of each element; it is a commonly used physical quantity for outline characterization in this field and will not be described in detail here. The outline changes of graphics and non-graphic elements (such as text) differ significantly. Generally, the outline changes of graphic elements are relatively gradual, while the outline changes of non-graphic elements are more rapid. Analyzing the outline vector information determines the curvature change characteristics of the boundary outline of each element, thereby quantifying the difference in curvature change per unit length of the boundary outline of each element. If the curvature change difference of an element's boundary contour within a unit length is less than a preset difference threshold, the element is determined to be a graphic element; otherwise, the element is determined to be a non-graphic element, facilitating subsequent text character-level recognition only for non-graphic elements. Furthermore, non-graphic elements in drawings are used to identify and interpret graphic elements; that is, there is a correlation between non-graphic elements and graphic elements. Correspondingly, the positional relationship between non-graphic elements and their associated graphic elements on the drawing is relatively close, and all non-graphic elements associated with the same graphic element are usually concentrated within a certain directional range corresponding to the same graphic element. To centrally organize and classify all non-graphic elements associated with the same graphic element, based on the boundary contour position of the graphic element, the relative directional distribution information of each non-graphic element to the boundary contour of the graphic element is determined. This divides all non-graphic elements into several non-graphic element sets, ensuring that all non-graphic elements within the same set are concentrated within the aforementioned directional range corresponding to the graphic element, while also guaranteeing the correlation between all non-graphic elements within the same set.

[0131] In another embodiment, based on the boundary contour curvature variation characteristics, the elements are distinguished into graphic elements or non-graphic elements, including:

[0132] Extract the difference in curvature change of the boundary profile per unit length from the boundary profile curvature change characteristics;

[0133] Using 50 pixels as the reference length for the engineering drawing, the curvature change difference of the boundary contour within each unit length is standardized using the reference length to obtain the standardized value corresponding to the curvature change difference of the boundary contour within each unit length.

[0134] The standardized value corresponding to the curvature change difference of the boundary contour within each unit length is obtained by the following formula:

[0135]

[0136] Where ΔC represents the standardized value corresponding to the difference in curvature of the boundary profile per unit length; ΔC x L represents the curvature variation difference of the boundary profile per unit length; L represents the actual total length of the boundary profile; L0 represents the reference length of the engineering drawing; C max This represents the maximum theoretically possible difference in curvature variation in engineering drawings.

[0137] The standard average value of the curvature change difference of the boundary profile is obtained by using the standardized value corresponding to the curvature change difference of the boundary profile within each unit length.

[0138] The standardized values ​​corresponding to the curvature change difference of the boundary contour within each unit length are compared with the preset standardized thresholds. The boundary contour positions with standardized values ​​not lower than the preset standardized thresholds are selected as high-band positions.

[0139] The number of high-band positions and the standardized values ​​corresponding to the curvature change difference of each high-band position are retrieved;

[0140] The high-band influence factor is obtained by using the number of high-band locations and the standardized value corresponding to the curvature change difference of each high-band location.

[0141] The high-band impact factor is obtained using the following formula:

[0142]

[0143] Where R represents the high-band influence factor; n represents the number of high-band locations; P represents the percentage of high-band locations relative to the total number of locations per unit length of the boundary profile; ΔC i ΔC represents the standardized value corresponding to the curvature change difference at the i-th high-band position; p This represents the standard mean of the difference in curvature variation;

[0144] The comprehensive evaluation coefficient of curvature fluctuation is obtained by combining the high-band influence factor with the standard average value of curvature change difference.

[0145] The comprehensive evaluation coefficient of curvature fluctuation is obtained by the following formula:

[0146]

[0147] Where K represents the comprehensive evaluation coefficient of curvature fluctuation; R represents the high-band influence factor; ΔC p This represents the standard mean of the difference in curvature variation;

[0148] The comprehensive evaluation coefficient of curvature fluctuation is compared with a preset coefficient threshold. If the comprehensive evaluation coefficient of curvature fluctuation exceeds the preset coefficient threshold, it is determined to be a non-graphic element; otherwise, it is determined to be a graphic element.

[0149] The beneficial effects of the above embodiments are achieved by eliminating interference from elements of different sizes and extreme curvature values ​​through the coordinated standardization of the difference between the baseline length, actual total length, and theoretical maximum curvature change. This unifies the evaluation criteria, enabling graphic and non-graphic elements of different specifications to have comparable curvature feature representations, thus improving the fairness and accuracy of classification. The standard average value of the curvature change difference comprehensively reflects the overall level of curvature change in the element boundary contour. The high-band influence factor accurately characterizes the local concentration characteristics of curvature fluctuations by integrating the proportion of high-band elements and the deviation of the standardized value of a single high-band element relative to the average value. The dual-dimensional features synergistically cover both overall and local curvature features, overcoming the limitations of single feature representation. The comprehensive evaluation coefficient of curvature fluctuation integrates the high-band influence factor and the standard average value of the curvature change difference, amplifying the feature differences between graphic elements (gentle curvature, few high fluctuations) and non-graphic elements (drastic curvature, concentrated high fluctuations), significantly improving the distinguishability between the two types of elements and reducing the risk of misjudging elements with ambiguous boundaries. Simultaneously, the above technical solutions effectively improve the stability and reliability of classification results. Furthermore, the technical solution is adapted to the element characteristics of architectural and engineering drawings, specifically capturing the high curvature fluctuation characteristics of non-graphical elements such as text annotations and the gentle curvature characteristics of graphic elements. The classification logic closely matches the actual element attributes of the drawings, significantly improving the adaptability and accuracy of element classification in engineering drawing scenarios. All calculation processes are based on basic arithmetic operations and statistical analysis, without complex algorithms or iterative processes. In engineering applications, it can be quickly implemented using conventional image processing modules, balancing classification accuracy and implementation efficiency, and possessing strong engineering practicality and scalability.

[0150] In another embodiment, the string generation module is used to group all non-graphical elements based on the connected components of all non-graphical elements under the non-graphical element set, and generate several text strings corresponding to the non-graphical element set, including:

[0151] Adaptive color reduction processing is performed on all non-graphic elements under the non-graphic element set to extract the connected components of all non-graphic elements; based on the connected components, spatial adjacency recognition is performed on all non-graphic elements to group all non-graphic elements into several text strings corresponding to the non-graphic element set.

[0152] The text result determination module is used to perform fuzzy classification and recognition on text strings to obtain text results for non-graphical element sets, including:

[0153] The text string is fuzzy classified and identified to obtain the text information contained in the text string. The text information is then logically corrected to obtain the text result of the non-graphic element set.

[0154] The beneficial effects of the above embodiments are that the correlation between all non-graphical elements under the non-graphical element set is not the same. There are corresponding connected components between any two non-graphical elements, and the connectivity of these components is related to the text strings to which the two non-graphical elements belong. Generally speaking, the connected components of two non-graphical elements belonging to the same text string are significantly different from those of two non-graphical elements belonging to different text strings. Therefore, adaptive color reduction processing is performed on all non-graphical elements under the non-graphical element set to extract the connected components between all non-graphical elements. These connected components are then analyzed to determine the spatial adjacency relationship (e.g., spatial spacing) between any two non-graphical elements. This grouping of all non-graphical elements into several text strings corresponding to the non-graphical element set ensures that each text string completely contains the corresponding text characters. Furthermore, fuzzy classification and recognition are performed on the text strings to obtain the text information contained within them. Semantic and word order logic corrections are then applied to this text information to obtain the final text result of the non-graphical element set, ensuring the accuracy of the text result.

[0155] In another embodiment, the text result preprocessing and adding module is used to preprocess the text results based on the spatial distribution information of graphic and non-graphic elements on the drawing, and then overlay and add them to the corresponding areas of the drawing, including:

[0156] The spatial distribution of graphic elements and non-graphic elements is identified within the global scope of the drawing to obtain the spatial occupancy information of each graphic element and non-graphic element.

[0157] Based on the spatial occupancy information of graphic and non-graphic elements, the element size ratio between graphic and non-graphic elements is determined, and the font size of the text result is preprocessed accordingly. Then, based on the spatial occupancy information of non-graphic elements, the preprocessed text result is added to the corresponding area of ​​the drawing.

[0158] The beneficial effects of the above embodiments are that all graphic elements and all non-graphic elements in a drawing occupy a certain proportion of space within the global area of ​​the drawing. If the corresponding area occupied by the text result after being re-added to the drawing differs significantly from the original area occupied by the non-graphic elements, or if the proportion of the corresponding area occupied by the text result after being re-added to the drawing differs significantly from the original proportion, the visual effect of the text result in the drawing will be reduced, resulting in visual disharmony between the original graphic elements and the text result. Therefore, spatial distribution identification of graphic elements and non-graphic elements within the global area of ​​the drawing is performed to obtain the spatial occupancy information of each graphic element and non-graphic element. Based on the spatial occupancy information of graphic elements and non-graphic elements, the element size ratio between graphic elements and non-graphic elements is determined. This is used to preprocess the text result by adjusting its font size. Based on the spatial occupancy information of non-graphic elements, the preprocessed text result is then overlaid onto the corresponding area of ​​the drawing, thereby maintaining visual harmony between the text result overlaid on the drawing and the original graphic elements on the drawing.

[0159] In summary, this scene-based text recognition-based drawing processing method and system identifies all graphic and non-graphic elements within the global scope of a drawing. Based on the shape features of graphic elements, it divides the drawing into several non-graphic element sets. Based on the connected components of all non-graphic elements within each set, it generates several text strings corresponding to those sets and performs fuzzy classification to identify the text results for each set. Based on the spatial distribution information of graphic and non-graphic elements on the drawing, it preprocesses the text results and overlays them onto the corresponding areas of the drawing, distinguishing between non-graphic and graphic elements. It then performs connectivity recognition on the non-graphic elements, initially labeling the text strings within them to obtain their corresponding text results. This ensures that the text results accurately and completely represent the meaning of the text, achieving synchronous association and recognition of graphics and text within the drawing, improving the efficiency and accuracy of drawing editing, and enabling correct recognition of text content across the entire drawing scope.

[0160] The above is only one specific embodiment of the present invention, and any improvements made based on the concept of the present invention shall be considered within the scope of protection of the present invention.

Claims

1. A drawing processing method based on scene text recognition, characterized in that, include: The drawing is identified to obtain all graphic elements and all non-graphic elements within the global scope of the drawing. Based on the shape characteristics of the graphic elements, all non-graphic elements are divided into several sets of non-graphic elements. Based on the connected components of all non-graphical elements under the non-graphical element set, all non-graphical elements are grouped to generate several text strings corresponding to the non-graphical element set. The text string is subjected to fuzzy classification and recognition to obtain the text result of the non-graphic element set; Based on the spatial distribution information of the graphic elements and non-graphic elements on the drawing, the text results are preprocessed in shape and then overlaid onto the corresponding area of ​​the drawing.

2. The drawing processing method based on scene text recognition as described in claim 1, characterized in that: The drawing is identified to obtain all graphic elements and all non-graphic elements within the global scope of the drawing. Based on the shape characteristics of the graphic elements, all non-graphic elements are divided into several sets of non-graphic elements, including: Contour recognition is performed on the drawing to obtain the contour vector information of each element within the global scope of the drawing; based on the contour vector information, the boundary contour curvature change characteristics of each element are determined; wherein, the boundary contour curvature change characteristics refer to the difference in curvature change of the boundary contour of the element within a unit length; Based on the boundary contour curvature variation characteristics, the elements are classified into graphic elements or non-graphic elements; Based on the boundary contour position of the graphic element, determine the relative orientation distribution information of each non-graphic element with respect to the boundary contour of the graphic element; based on the relative orientation distribution information, divide all non-graphic elements into several non-graphic element sets.

3. The drawing processing method based on scene text recognition as described in claim 2, characterized in that: Based on the aforementioned boundary contour curvature variation characteristics, the elements are classified into graphic elements or non-graphic elements, including: Extract the difference in curvature change of the boundary profile per unit length from the boundary profile curvature change characteristics; Using 50 pixels as the reference length for the engineering drawing, the curvature change difference of the boundary contour within each unit length is standardized using the reference length to obtain the standardized value corresponding to the curvature change difference of the boundary contour within each unit length. The standardized value corresponding to the curvature change difference of the boundary contour within each unit length is obtained by the following formula: Where ΔC represents the standardized value corresponding to the difference in curvature of the boundary profile per unit length; ΔC x L represents the curvature variation difference of the boundary profile per unit length; L represents the actual total length of the boundary profile; L0 represents the reference length of the engineering drawing; C max This represents the maximum theoretically possible difference in curvature variation in engineering drawings. The standard average value of the curvature change difference of the boundary profile is obtained by using the standardized value corresponding to the curvature change difference of the boundary profile within each unit length. The standardized values ​​corresponding to the curvature change difference of the boundary contour within each unit length are compared with the preset standardized thresholds. The boundary contour positions with standardized values ​​not lower than the preset standardized thresholds are selected as high-band positions. The number of high-band positions and the standardized values ​​corresponding to the curvature change difference of each high-band position are retrieved; The high-band influence factor is obtained by using the number of high-band locations and the standardized value corresponding to the curvature change difference of each high-band location. The high-band impact factor is obtained using the following formula: Where R represents the high-band influence factor; n represents the number of high-band locations; P represents the percentage of high-band locations relative to the total number of locations per unit length of the boundary profile; ΔC i ΔC represents the standardized value corresponding to the curvature change difference at the i-th high-band position; p This represents the standard mean of the difference in curvature variation; The comprehensive evaluation coefficient of curvature fluctuation is obtained by combining the high-band influence factor with the standard average value of curvature change difference. The comprehensive evaluation coefficient of curvature fluctuation is obtained by the following formula: Where K represents the comprehensive evaluation coefficient of curvature fluctuation; R represents the high-band influence factor; ΔC p This represents the standard mean of the difference in curvature variation; The comprehensive evaluation coefficient of curvature fluctuation is compared with a preset coefficient threshold. If the comprehensive evaluation coefficient of curvature fluctuation exceeds the preset coefficient threshold, it is determined to be a non-graphic element; otherwise, it is determined to be a graphic element.

4. The drawing processing method based on scene text recognition as described in claim 1, characterized in that: Based on the connected components of all non-graphical elements under the non-graphical element set, all non-graphical elements are grouped to generate several text strings corresponding to the non-graphical element set. Performing fuzzy classification and recognition on the text string to obtain the text results of the non-graphic element set includes: Adaptive color reduction processing is performed on all non-graphic elements under the non-graphic element set to extract the connected components of all non-graphic elements; based on the connected components, spatial adjacency recognition is performed on all non-graphic elements to group all non-graphic elements into several text strings corresponding to the non-graphic element set. The text string is subjected to fuzzy classification and recognition to obtain the text information contained in the text string, and the text information is subjected to logical correction processing to obtain the text result of the non-graphic element set.

5. The drawing processing method based on scene text recognition as described in claim 1, characterized in that: Based on the spatial distribution information of the graphic elements and non-graphic elements on the drawing, the text result is preprocessed in terms of shape and then overlaid onto the corresponding area of ​​the drawing, including: The spatial distribution of the graphic elements and the non-graphic elements is identified within the global scope of the drawing to obtain the spatial occupancy information of each graphic element and the non-graphic element. Based on the spatial occupancy information of the graphic elements and the non-graphic elements, the element size ratio between the graphic elements and the non-graphic elements is determined, and the font size of the text result is preprocessed accordingly; then, based on the spatial occupancy information of the non-graphic elements, the preprocessed text result is overlaid and added to the corresponding area of ​​the drawing.

6. A drawing processing system based on scene text recognition, characterized in that, include: The drawing recognition module is used to recognize the drawing and obtain all graphic elements and all non-graphic elements within the global scope of the drawing; The element division module is used to divide all non-graphic elements into several non-graphic element sets based on the shape characteristics of the graphic elements. The string generation module is used to group all non-graphic elements based on the connected components of all non-graphic elements under the non-graphic element set, and generate several text strings corresponding to the non-graphic element set. The text result determination module is used to perform fuzzy classification and recognition on the text string to obtain the text result of the non-graphic element set; The text result preprocessing and adding module is used to preprocess the text result in shape and then overlay it onto the corresponding area of ​​the drawing based on the spatial distribution information of the graphic elements and the non-graphic elements on the drawing.

7. The drawing processing system based on scene text recognition as described in claim 6, characterized in that: The drawing recognition module is used to recognize the drawing and obtain all graphic elements and all non-graphic elements within the global scope of the drawing, including: Contour recognition is performed on the drawing to obtain the contour vector information of each element within the global scope of the drawing; based on the contour vector information, the boundary contour curvature change characteristics of each element are determined; wherein, the boundary contour curvature change characteristics refer to the difference in curvature change of the boundary contour of the element within a unit length; Based on the boundary contour curvature variation characteristics, the elements are classified into graphic elements or non-graphic elements; The element partitioning module is used to divide all non-graphical elements into several non-graphical element sets based on the shape characteristics of the graphic elements, including: Based on the boundary contour position of the graphic element, determine the relative orientation distribution information of each non-graphic element with respect to the boundary contour of the graphic element; based on the relative orientation distribution information, divide all non-graphic elements into several non-graphic element sets.

8. The drawing processing system based on scene text recognition as described in claim 7, characterized in that: Based on the aforementioned boundary contour curvature variation characteristics, the elements are classified into graphic elements or non-graphic elements, including: Extract the difference in curvature change of the boundary profile per unit length from the boundary profile curvature change characteristics; Using 50 pixels as the reference length for the engineering drawing, the curvature change difference of the boundary contour within each unit length is standardized using the reference length to obtain the standardized value corresponding to the curvature change difference of the boundary contour within each unit length. The standardized value corresponding to the curvature change difference of the boundary contour within each unit length is obtained by the following formula: Where ΔC represents the standardized value corresponding to the difference in curvature of the boundary profile per unit length; ΔC x L represents the curvature variation difference of the boundary profile per unit length; L represents the actual total length of the boundary profile; L0 represents the reference length of the engineering drawing; C max This represents the maximum theoretically possible difference in curvature variation in engineering drawings. The standard average value of the curvature change difference of the boundary profile is obtained by using the standardized value corresponding to the curvature change difference of the boundary profile within each unit length. The standardized values ​​corresponding to the curvature change difference of the boundary contour within each unit length are compared with the preset standardized thresholds. The boundary contour positions with standardized values ​​not lower than the preset standardized thresholds are selected as high-band positions. The number of high-band positions and the standardized values ​​corresponding to the curvature change difference of each high-band position are retrieved; The high-band influence factor is obtained by using the number of high-band locations and the standardized value corresponding to the curvature change difference of each high-band location. The high-band impact factor is obtained using the following formula: Where R represents the high-band influence factor; n represents the number of high-band locations; P represents the percentage of high-band locations relative to the total number of locations per unit length of the boundary profile; ΔC i ΔC represents the standardized value corresponding to the curvature change difference at the i-th high-band position; p This represents the standard mean of the difference in curvature variation; The comprehensive evaluation coefficient of curvature fluctuation is obtained by combining the high-band influence factor with the standard average value of curvature change difference. The comprehensive evaluation coefficient of curvature fluctuation is obtained by the following formula: Where K represents the comprehensive evaluation coefficient of curvature fluctuation; R represents the high-band influence factor; ΔC p This represents the standard mean of the difference in curvature variation; The comprehensive evaluation coefficient of curvature fluctuation is compared with a preset coefficient threshold. If the comprehensive evaluation coefficient of curvature fluctuation exceeds the preset coefficient threshold, it is determined to be a non-graphic element; otherwise, it is determined to be a graphic element.

9. The drawing processing system based on scene text recognition as described in claim 6, characterized in that: The string generation module is used to group all non-graphical elements based on the connected components of all non-graphical elements under the non-graphical element set, and generate several text strings corresponding to the non-graphical element set, including: Adaptive color reduction processing is performed on all non-graphic elements under the non-graphic element set to extract the connected components of all non-graphic elements; based on the connected components, spatial adjacency recognition is performed on all non-graphic elements to group all non-graphic elements into several text strings corresponding to the non-graphic element set. The text result determination module is used to perform fuzzy classification and recognition on the text string to obtain the text result of the non-graphic element set, including: The text string is subjected to fuzzy classification and recognition to obtain the text information contained in the text string, and the text information is subjected to logical correction processing to obtain the text result of the non-graphic element set.

10. The drawing processing system based on scene text recognition as described in claim 6, characterized in that: The text result preprocessing and adding module is used to preprocess the text results based on the spatial distribution information of the graphic elements and non-graphic elements on the drawing, and then overlay and add them to the corresponding areas of the drawing, including: The spatial distribution of the graphic elements and the non-graphic elements is identified within the global scope of the drawing to obtain the spatial occupancy information of each graphic element and the non-graphic element. Based on the spatial occupancy information of the graphic elements and the non-graphic elements, the element size ratio between the graphic elements and the non-graphic elements is determined, and the font size of the text result is preprocessed accordingly; then, based on the spatial occupancy information of the non-graphic elements, the preprocessed text result is overlaid and added to the corresponding area of ​​the drawing.

Images