Calligraphy practice auxiliary system and method based on intelligent visual system

Abstract

The invention discloses a calligraphy practice auxiliary system and method based on an intelligent visual system, and the method comprises the steps: firstly, synchronously collecting the three-dimensional track, wielding angle and pressure distribution data of a pen point through a high-speed camera and an inertial measurement unit, comprehensively recording the actions of a real pen and a virtual pen, and solving the defect that the traditional system is difficult to capture the virtual pen in the air; secondly, key indexes such as potential energy density and rhythm entropy are quantified through space-time sequence modeling, and dynamic coherence and rhythm changes of cursive scripts are accurately analyzed; and thirdly, an ideal virtual pen path is generated based on the light-weight GAN network, and the deviation degree is calculated, so that a more intuitive improvement basis is provided for a user. And finally, aligning the speed-pressure distribution matrixes of the user and the famous family by means of a dynamic time warping algorithm, quantitatively evaluating the charm coherence, projecting a three-dimensional track and a charm thermodynamic diagram in real time, and providing personalized training suggestions. According to the method, the limitation that a traditional calligraphy auxiliary technology depends on static trajectory analysis is broken through, and refined capture and quantitative evaluation of cursive writing dynamic writing are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of data recognition, particularly to the technical field of reading or recognizing written characters, and specifically to a calligraphy practice assistance system and method based on an intelligent vision system. Background Art

[0002] Current vision-based calligraphy assistance systems mainly use image comparison and pen-tip trajectory tracking technologies to provide automated guidance for regular-structured fonts such as regular script and official script. These systems rely on standardized stroke templates and static image analysis, and evaluate writing accuracy by capturing the physical trajectory of paper contact. With the in-depth application of artificial intelligence in the art field, calligraphy assistance technology has gradually expanded from basic glyph correction to the level of pen movement analysis. However, cursive script, especially wild cursive script, as the peak form of calligraphy art, has much higher dynamic coherence and personalized expression than regular fonts. Wild cursive script is famous for "writing in one stroke", emphasizing the abstract beauty of flowing qi and connecting meanings even when the pen strokes are disjointed. The writing process involves a large number of virtual strokes in the air and rhythm changes. The traditional technology system based on static trajectory or single-character structure analysis is difficult to adapt to the highly dynamic artistic creation requirements.

[0003] Although existing calligraphy practice assistance systems have improved the practice efficiency to a certain extent, there are still significant deficiencies in the practice assistance of cursive script, especially wild cursive script. Traditional systems are mostly based on static image comparison or discrete trajectory point analysis, and it is difficult to capture the core characteristics of dynamic cursive writing. For example, they cannot effectively quantify dynamic elements such as writing speed, strength, and rhythm, let alone evaluate the coherence between brushstrokes and the vividness of the overall qi charm. In addition, existing systems often focus on the standardization of stroke structures and ignore the crucial personalized expression in calligraphy art. For a calligraphy style like wild cursive script that emphasizes improvisation and emotional expression, the mechanical feedback of existing systems often goes against the essence of calligraphy art. More seriously, the existing technical framework lacks support for special writing techniques such as "connecting meanings even when the pen strokes are disjointed" in cursive script, and it is difficult to capture the trajectory of the pen tip connecting virtually in the air, resulting in a significant reduction in the assistance effect.

[0004] Therefore, it is necessary to improve the calligraphy practice assistance system and method in the prior art to solve the above problems. Summary of the Invention

[0005] The present invention overcomes the deficiencies of the prior art and provides a calligraphy practice assistance system and method based on an intelligent vision system, aiming to solve the problems of insufficient dynamic capture and ignored personalized expression for cursive script in the prior art.

[0006] To achieve the above object, the technical solution adopted by the present invention is: A calligraphy practice method based on an intelligent vision system, comprising: S1. Synchronously collect the aerial trajectory, pen movement angle, and pressure distribution data during writing through high-speed vision and an inertial measurement unit; S2. Use spatio-temporal sequence modeling to extract dynamic features from the continuous pen movement trajectory, and quantify the potential energy density and rhythm entropy; S3. Use a lightweight GAN network to generate an ideal virtual pen path, calculate the distance between the user's actual virtual pen and the generated path, and quantify the user's virtual pen deviation; S4. Align the speed-pressure distribution matrices of the user and famous calligraphers' works using the dynamic time warping algorithm to evaluate the user's rhythm coherence; S5. Project the three-dimensional trajectory and rhythm heat map, and generate rhythm correction and special training suggestions in real time.

[0007] In a preferred embodiment of the present invention, a high-speed camera is paired with an 800 - 850nm infrared filter to track the hydrophobic reflective marker points of the pen tip, and synchronously collect the three-dimensional coordinates, speed, acceleration, pen movement angle, and pressure data of the pen tip; among them, the main camera vertically looks down on the paper surface at 40 - 45cm to capture the XY plane trajectory, and the side-view camera captures the Z-axis suspension height at an inclination angle of 30 - 60° and a distance of 20 - 30cm from the pen tip; a nine-axis IMU is set at the center of gravity of the pen barrel to monitor acceleration and angular velocity, and a micro pressure sensor with a measuring range of 0 - 10N is set on the lower side of the pen barrel holding area to record the pressure distribution.

[0008] In a preferred embodiment of the present invention, the dynamic feature extraction in step S2 includes: Process the spatio-temporal sequence through a Transformer encoder ; Output the hidden state sequence , which is the motion feature at time t and its correlation with the strokes before and after; The quantification indexes include: The potential energy density measures the intensity of dynamic changes, and the potential energy density , where is the inclination change amount per unit time, is the inclination change threshold, and T is a certain period of time; Discretize the speed sequence into K intervals; the rhythm entropy is , where is the speed value and

[0009] is the probability of its occurrence. In a preferred embodiment of the present invention, in step S3, the generator of the lightweight GAN network takes the current actual pen end coordinate , and the next actual pen start coordinate ; The discriminator inputs the actual virtual stroke path of the user or the generated path , and outputs the probability value that the path conforms to the cursive calligraphy rules; the virtual stroke deviation degree is calculated by using the dynamic time warping algorithm to align the user path with the ideal path , , where is the optimal alignment path, is the Euclidean distance; the deviation degree is quantified , is the total length of the ideal path

[0010] In a preferred embodiment of the present invention, in step S4, the speed sequence extracted from step S2 and the pressure sequence ; Combine speed and pressure , is the adjustment parameter; Calculate the frame - by - frame Euclidean distance between the user matrix and the master matrix ; i is the user time - frame index, , j is the master time - frame index, ; Qi - yun coherence evaluation: Quantify the rhythm matching degree , the larger the value, the higher the rhythm similarity, is the attenuation coefficient; Similarity score , , is the ratio of the user's potential energy density to that of the master, is the adjustment factor

[0011] In a preferred embodiment of the present invention, in step S5, the three - dimensional trajectory is projected in real - time through an AR device or a screen. The real - stroke trajectory is displayed in red, the user's virtual - stroke trajectory is displayed in semi - transparent white, and the ideal virtual - stroke path is marked in blue

[0012] The present invention provides a calligraphy practice device based on an intelligent vision system, including: A high - speed camera module, including a main camera and a side - view camera; the high - speed camera module is connected to the data processing module The nib marking module, the inertial measurement unit module, the pressure sensing module, and the data transmission unit are all provided on the writing brush; the nib marking module includes a hydrophobic reflective marking point fixed 3 mm from the nib; the inertial measurement unit module includes a nine-axis IMU provided at the center of gravity of the pen barrel; the pressure sensing module includes a 0-10N micro pressure sensor on the lower side of the pen barrel grip area; the data transmission unit includes a Bluetooth module at the pen tip; The Bluetooth module is connected to the data processing module to transmit data to the display screen.

[0013] In a preferred embodiment of the present invention, the main camera and the side camera are high-speed cameras above 120 fps, and are equipped with an 800-850 nm infrared filter; the main camera vertically looks down on the paper surface at 40-45 cm, and the side camera is installed at an inclination angle of 30-60°, 20-30 cm away from the nib.

[0014] In a preferred embodiment of the present invention, the structure of the hydrophobic reflective marking point from bottom to top includes: a polyimide film base layer with a thickness of 0.05-0.07 mm, a UV-cured optical glue bonding layer with a thickness of 30-50 μm, a fluorinated silicon dioxide nano-coating with a thickness of 20-30 μm, a contact angle >150°, used to block the infiltration of ink, and ceramic microbeads with a diameter of 0.1-0.12 mm. The thickness of the hydrophobic reflective marking point is less than 0.2 mm, and the weight <0.02 g.

[0015] The present invention provides a calligraphy practice assistance system based on an intelligent vision system, including: A multi-modal data acquisition module for collecting data on the three-dimensional aerial trajectory of the nib, the writing angle, the pressure distribution, and the paper surface contact trajectory; A dynamic feature extraction module for quantifying key dynamic indicators and identifying real / virtual strokes, pen tip hovering, and virtual connection actions; A virtual stroke path generation module for generating an ideal virtual stroke path that conforms to the cursive script rules through a lightweight GAN network; A virtual stroke path evaluation module for calculating the deviation degree between the user's actual virtual stroke and the ideal path; An artistic conception coherence evaluation module for quantifying the rhythm matching degree and the artistic conception coherence; A real-time feedback module for projecting the three-dimensional trajectory in real time and generating a personalized report.

[0016] The present invention solves the defects existing in the background technology, and the present invention has the following beneficial effects: (1) The present invention proposes a calligraphy practice assistance system and method based on an intelligent vision system, mainly targeting the practice needs of complex calligraphy forms such as wild cursive. First, the three-dimensional trajectory of the pen tip, the writing angle, and the pressure distribution data are synchronously collected through a high-speed camera and an inertial measurement unit to comprehensively record the real pen and virtual pen movements, solving the defect that traditional systems are difficult to capture virtual pens in the air. Second, spatio-temporal sequence modeling is used to quantify key indicators such as potential energy density and rhythm entropy to accurately analyze the dynamic coherence and rhythm changes of cursive. Third, an ideal virtual pen path is generated based on a lightweight GAN network and the deviation is calculated to provide a more intuitive improvement basis for users. Finally, the dynamic time warping algorithm is used to align the speed-pressure distribution matrices of the user and the master calligrapher, quantitatively evaluate the coherence of qi and rhythm, and project the three-dimensional trajectory and the qi and rhythm heat map in real time to provide personalized training suggestions. Overall, the invention breaks through the limitation of traditional calligraphy assistance technology relying on static trajectory analysis, realizes the refined capture, quantitative evaluation, and real-time feedback of dynamic cursive writing, helps users deeply understand and master the qi and rhythm flow and personalized expression techniques of wild cursive, and effectively improves the efficiency of calligraphy practice and the artistic perception ability.

[0017] (2) The present invention uses spatio-temporal sequence modeling to continuously model the pen movement trajectory and extract dynamic features. The Transformer encoder is used to process the spatio-temporal sequence and output a hidden state sequence, and then the potential energy density and rhythm entropy are quantified. This combination realizes the accurate quantification of the dynamic change intensity and rhythm changes of cursive writing. Compared with the prior art, it further achieves the in-depth analysis effect of the coherence of qi and rhythm and the dynamic beauty of cursive, and can effectively identify key actions such as real pen / virtual pen, pen tip hovering, and virtual connection.

[0018] (3) The present invention combines a lightweight GAN network and a dynamic time warping algorithm to directly generate an ideal virtual pen path and quantify the deviation of the user's virtual pen. The generator outputs a sequence of ideal virtual pen path points according to the coordinates of the current real pen end point and the next real pen start point, and the discriminator outputs the probability value that the path conforms to the cursive writing rules. Then, the dynamic time warping algorithm is used to align the user's path with the ideal path and calculate the deviation. This combination realizes the accurate evaluation and visual comparison of the user's virtual pen trajectory. Compared with the prior art, it further achieves the accurate diagnosis and dynamic correction effect of the virtual pen artistic effect in wild cursive calligraphy, which helps users improve their writing skills.

[0019] (4) By combining high-speed vision with an inertial measurement unit, the present invention directly analyzes and synchronously collects data on the aerial trajectory, pen movement angle, and pressure distribution during writing. The main camera vertically looks down on the paper surface at a distance of 40 - 45 cm to capture the XY-plane trajectory, and the side-view camera captures the Z-axis suspension height at an inclination angle of 30 - 60° and a distance of 20 - 30 cm from the pen tip. Meanwhile, a nine-axis IMU is set at the center of gravity of the pen shaft to monitor acceleration and angular velocity, and a micro pressure sensor is set on the lower side of the pen-holding area to record the pressure distribution. This combination enables all-round data collection during the cursive writing process, including aerial virtual pen movements and dynamic characteristics. Compared with the prior art, it further achieves the precise capture effect of special writing techniques such as "connected in spirit but broken in form" in complex calligraphy forms like wild cursive script.

[0020] (5) By combining speed-pressure distribution matrix analysis and rhythm matching degree quantification, the present invention directly evaluates the coherence of the user's and masterpieces' artistic conceptions. Obtain the standardized speed-pressure matrix of the master from the cursive calligraphy composition knowledge base, combine speed and pressure to calculate the Euclidean distance between frames, and then quantify the rhythm matching degree and similarity score. This combination achieves the precise matching and quantitative evaluation of the user's writing rhythm and the rhythm of masterpieces. Compared with the prior art, it further achieves the scientific guidance effect on personalized expression and emotional expression in the creation of cursive calligraphy art, enabling users to better grasp the flow of artistic conception and abstract beauty of cursive script. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings; Figure 1 is a flowchart of a calligraphy practice method based on an intelligent vision system according to a preferred embodiment of the present invention; Figure 2 is a structural diagram of a calligraphy practice device based on an intelligent vision system according to a preferred embodiment of the present invention; In the figure: 1, main camera; 2, side-view camera; 3, hydrophobic reflective marking points; 4, nine-axis IMU; 5, Bluetooth unit; 6, screen; 7, rice paper; 8, writing brush. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0024] In the description of the present application, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the scope of protection of the present application.

[0025] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0026] Application Overview: Traditional cameras and stylus sensors only record the contact trajectory of the paper surface. The system relies on discrete data collection and cannot capture the key "virtual pen in the air" movements of cursive calligraphy, such as pen tip rotation and hovering connection, resulting in the complete neglect of the artistic effect of "the idea is connected but the pen is broken", resulting in the loss of key dynamic parameters such as the virtual pen path and lifting force, making it impossible for the system to build a complete "real pen-virtual pen" motion chain; the static feature extraction model is in conflict with the temporal dependency of cursive calligraphy. The beauty of cursive calligraphy comes from the implicit connection between strokes, while traditional algorithms only deconstruct isolated strokes and ignore the continuity of time and space. The core value of cursive calligraphy lies in the alternation of sudden and slow strokes and potential energy changes. The standardized template matching mechanism conflicts with the personalized expression of cursive calligraphy.

[0027] This application proposes improvements to these deep-seated defects. By introducing multimodal data collection and spatiotemporal modeling technology, this application aims to build a new auxiliary system that can dynamically capture the entire process of cursive writing, quantitatively evaluate the coherence of the spirit and rhyme, and support personalized artistic expression.

[0028] Exemplary methods: like Figure 1 As shown, a calligraphy practice method based on an intelligent visual system comprises the following steps: S1. Synchronously collect the aerial trajectory, pen movement angle, and pressure distribution data during writing through high-speed vision and an inertial measurement unit; S2. Use spatio-temporal sequence modeling to extract dynamic features from the continuous pen movement trajectory, and quantify the potential energy density and rhythm entropy; S3. Use a lightweight GAN network to generate an ideal virtual pen path, calculate the distance between the user's actual virtual pen and the generated path, and quantify the deviation of the user's virtual pen; S4. Adopt the dynamic time warping algorithm to align the speed-pressure distribution matrices of the user and the master's works, and evaluate the user's rhythm coherence; S5. Project the three-dimensional trajectory and rhythm heat map, and generate rhythm correction and special training suggestions in real time.

[0029] Traditional calligraphy assistance systems only record the trajectory of paper contact, completely ignoring the aerial pen movement in wild cursive script, such as the "virtual front turning" in Huaisu's "Autobiographical Account". These movements are the core carriers of the artistic effect of meaning connection with pen breaks. Their absence will lead to: the break of the dynamic chain, and the inability to construct a continuous movement model of real pen - virtual pen; the dynamic features such as pressing and lifting force, sudden and slow rhythm cannot be quantified; the system can only evaluate the static structure and cannot analyze the rhythm coherence.

[0030] Traditional cameras and sensors only record the trajectory of paper contact and cannot capture the key "aerial virtual pen" movements in wild cursive script, such as pen tip rotation and hovering connection, resulting in the complete neglect of the artistic effect of "meaning connection with pen breaks". In order to comprehensively capture the writing of cursive script, especially the "aerial virtual pen" movements and dynamic features of wild cursive script, in step S1, a high-speed camera above 120fps is used, and an 800 - 850nm infrared filter is used to enhance the tracking ability of the pen tip marking point. A hydrophobic reflective marking point is fixed 3mm away from the pen tip of the writing brush, avoiding the writing surface of the pen tip to ensure that it can still be tracked by the infrared lens after being soaked by ink. The high-speed camera is divided into: a main camera and a side-view camera. The main camera vertically looks down on the paper surface at 40 - 45cm to capture the XY plane trajectory, and the side-view camera is installed at an inclination angle of 30 - 60°, 20 - 30cm away from the pen tip, to capture the Z-axis suspension height. The high-speed camera covers the pen tip suspension area and the paper writing range; Fixing a hydrophobic reflective marking point 3mm away from the pen tip of the writing brush needs to solve three major problems: anti-ink infiltration, precise positioning, and minimizing writing interference. The structure of the hydrophobic reflective marking point from bottom to top includes: a polyimide film base layer with a thickness of 0.05 - 0.07mm, a UV-cured optical glue bonding layer with a thickness of 30 - 50μm, a fluorinated silica nanocoating with a thickness of 20 - 30μm, a contact angle > 150° for blocking ink infiltration, and ceramic microbeads with a diameter of 0.1 - 0.12mm. The thickness of the hydrophobic reflective marking point is less than 0.2mm, and the weight < 0.02g, and the interference to writing can be ignored; realizing the continuous recording of the aerial trajectory and the paper surface trajectory.

[0031] In addition, an inertial measurement unit nine-axis IMU is provided at the center of gravity of the pen barrel to monitor acceleration, angular velocity, etc. in real time. By inertial measurement, the change of the writing angle and the dynamic rhythm are quantified. A micro pressure sensor is provided on the lower side of the pen barrel holding area to record the pressure distribution of the finger on the pen barrel during writing and correlate it with the change of ink shade. The measuring range is 0 - 10N and the accuracy is ±0.1N. A Bluetooth unit is provided at the pen tail for wireless data transmission. The air trajectory parameters include: three-dimensional coordinates, speed, and acceleration, which capture the meaning-linking actions such as the hovering and virtual connection of the pen tip and construct a complete motion chain. The pen movement angle parameters include: pitch angle, yaw angle, and rotational angular velocity, which quantify the rotation of the pen tip and the pressing force. The pressure distribution is correlated with the ink shade. The real pen area has a solid ink trace, corresponding to high pressure, and the virtual pen area has a light or flying white ink trace, corresponding to low pressure.

[0032] In step S2, the dynamic features of cursive writing are extracted through spatio-temporal sequence modeling, and the key indicators are quantified to analyze the coherence of the artistic conception and rhythm.

[0033] In step S2, the three-dimensional coordinates of the pen tip are: , where are the positions on the x, y, and z axes at time t, and the speed is the instantaneous speed ; the acceleration is , and the angular velocity is , and the pressure sensor is standardized as: ; where P is the actually measured pressure at time t.

[0034] Based on the collected parameters, the actions are classified: Judgment of real and virtual pen strokes: , for a real pen stroke, it is necessary to satisfy both paper contact and effective pressure, and a virtual pen stroke includes air movement or very light contact with the paper; Judgment of pen tip hovering: , where the speed, acceleration, and angular velocity all approach zero, indicating that the pen tip is stationary in the air for a short time; Judgment of virtual connection action: , the distance between the end point of the virtual pen stroke and the starting point of the next real pen stroke is less than 5mm, reflecting the coherence of "broken pen but continuous meaning".

[0035] The ink shade is correlated with the pressure: in the real pen area , the ink trace is full, corresponding to high pressure; in the virtual pen area , the ink trace is light or has flying white, corresponding to low pressure; The inclination angle quantifies the pen tip state: for writing with the pen tip in the center, the pitch angle is , and for writing with the pen tip on the side, the yaw angle is and the angular velocity is .

[0036] Input spatio-temporal sequence: ; Use a Transformer encoder to process spatio-temporal sequences and capture the dependencies between strokes through the self-attention mechanism; Output the hidden state sequence , is the motion feature at time t and its relevance to the preceding and following strokes; The potential energy density measures the intensity of dynamic changes. The potential energy density , where is the change amount of the inclination angle per unit time, is the inclination angle change threshold, and T is a certain period of time; Discretize the velocity sequence into K intervals; the rhythm entropy is , where is the velocity value is the probability of occurrence; an entropy value > 3.0 is a high entropy value, indicating a drastic rhythm change, which is a characteristic of wild cursive; an entropy value < 1.5 is a low entropy value, indicating a uniform rhythm and a lack of artistic tension.

[0037] In step S2, extract the dynamic correlation features of the pen movement trajectory through spatio-temporal sequence modeling and quantify the key indicators. Based on the dynamic features and quantification results extracted in S2, in step S3, use a lightweight GAN network to generate an ideal virtual pen path that conforms to the rules of cursive calligraphy. By comparing the deviation between the user's actual virtual pen trajectory and the generated path, accurately diagnose the weak links of "connected in spirit but broken in form" and provide a basis for dynamic correction.

[0038] In step S3, receive the spatio-temporal feature sequence output in step S2, including the real pen end point, virtual pen start and end point coordinates, and dynamic parameters; Load the pre-trained knowledge base of cursive calligraphy rules, which is constructed based on the virtual pen path dataset of famous artists' wild cursive works; Lightweight GAN network: The generator inputs the current real pen end point coordinates , and the next real pen start point coordinates , and its structure is a 3-layer fully connected network, outputting the ideal virtual pen path point sequence ; The discriminator inputs the user's actual virtual pen path or the generated path , and outputs: the probability value that the path conforms to the rules of cursive calligraphy; minimize the Wasserstein distance between the generated path and the authentic works of famous artists to ensure that the path conforms to the artistic law of "connected in spirit but broken in form".

[0039] The virtual pen deviation is calculated using the dynamic time warping algorithm to align the user's path with the ideal path , , where is the optimal alignment path, is the Euclidean distance.

[0040] Deviation quantification , is the total length of the ideal path, and the deviation threshold is set as: , generate a three-dimensional visual comparison diagram of the virtual pen path, with the user path marked in red and the ideal path marked in blue, and output the deviation score and weak link location.

[0041] Step S3 generates an ideal virtual pen path and deviation quantification through lightweight GAN, converting the abstract artistic effect of cursive calligraphy into computable parameters, providing an accurate basis for dynamic correction, and promoting calligraphy assistance from structural imitation to artistic expression.

[0042] In step S4, the speed sequence extracted from step S2 and pressure series , obtain the standardized speed-pressure matrix of famous cursive calligraphers from the cursive calligraphy knowledge base ; Combining speed and pressure , To adjust the parameters; Calculate the inter-frame Euclidean distance between the user matrix and the master matrix ; i is the user time frame index, , j is the index of the famous time frame, ; Assessment of rhythmic coherence: Rhythm matching metric , the larger the value, the higher the rhythm similarity. is the attenuation coefficient; Similarity score , , is the ratio of the user's potential energy density to that of the famous artists, It is an adjustment factor. The larger the value, the closer the continuity of the artistic conception is to that of famous artists.

[0043] In step S5, the system projects the user's written three-dimensional trajectory onto an AR device or high-resolution screen in real time, with the real pen trajectory displayed in bright red, the user's actual virtual pen trajectory presented in translucent white, and the ideal virtual pen path generated by AI highlighted in blue. When the user's actual virtual pen trajectory overlaps with the ideal path by more than 70%, the system automatically marks the trajectory in green and gives positive feedback reinforcement.

[0044] It supports 360° rotation to observe the writing trajectory. Users can choose 1-5 times slow motion playback function. The system automatically marks the key virtual joints in the air and dynamically inserts famous writing demonstration clips during the playback process, using arrows to instruct users to pay attention to the key nodes of virtual-real conversion.

[0045] The system collects users' practice data in real time and updates the GAN generation model once a week through reinforcement learning algorithms, making the ideal virtual stroke path more in line with the users' personal writing habits and progress curve. Core indicators such as the qi rhythm coherence score and the rationality of the virtual stroke trajectory are recorded for each practice. A weekly progress report is generated, showing the change trends of various indicators through line charts and marking key progress nodes.

[0046] Exemplary system: A calligraphy practice assistance system based on an intelligent vision system, comprising modules: A multi-modal data acquisition module for collecting data on the three-dimensional aerial trajectory of the pen tip, the pen movement angle, the pressure distribution, and the paper contact trajectory; A dynamic feature extraction module for quantifying key dynamic indicators and identifying real / virtual strokes, pen tip hovering, and virtual connection actions; A virtual stroke path generation module for generating an ideal virtual stroke path that conforms to the cursive script rules through a lightweight GAN network; A virtual stroke path evaluation module for calculating the deviation degree between the user's actual virtual stroke and the ideal path; A qi rhythm coherence evaluation module for quantifying the rhythm matching degree and the qi rhythm coherence; A real-time feedback module for projecting the three-dimensional trajectory in real time and generating a personalized report.

[0047] Exemplary device: As Figure 2 shown, a calligraphy practice device based on an intelligent vision system, comprising: The high-speed camera module includes: a main camera and a side-view camera. The main camera is vertically looking down at the paper surface at a distance of 40 - 45 cm, and is used to capture the writing trajectory of the pen tip on the XY plane; the side-view camera is installed at an inclination angle of 30 - 60°, and is 20 - 30 cm away from the pen tip, and is used to capture the suspended height of the pen tip in the Z-axis direction, so as to realize the capture of the three-dimensional space movement of the pen tip and provide data support for recording the aerial virtual stroke actions during the writing process.

[0048] A pen tip marking module fixes a hydrophobic reflective marking point 3 mm from the tip of the writing brush. This marking point is composed of a polyimide film base layer, a UV-cured optical glue bonding layer, a fluorinated silicon dioxide nano-coating, and ceramic microbeads, and has the characteristics of anti-ink infiltration, precise positioning, and little interference with writing, ensuring that during the writing process, the high-speed camera module can accurately track the position of the pen tip. Even after the ink soaks the pen tip, it can still be tracked by the infrared lens, realizing the continuous recording of the aerial trajectory and the paper surface trajectory.

[0049] The inertial measurement unit module, which is set at the center of gravity of the pen, is a nine-axis IMU that can monitor the acceleration, angular velocity and other motion data of the pen in real time, thereby quantifying the changes in pen angles and dynamic rhythms, and providing data basis for analyzing dynamic characteristics such as pen tip rotation and lifting and pressing force during the writing process.

[0050] The pressure sensing module is a miniature pressure sensor installed on the lower side of the pen holding area. It is used to record the pressure distribution of fingers on the pen when writing. Its measuring range is 0-10N and its accuracy is ±0.1N. It can associate the pressure data with the changes in the virtuality and reality of the ink. That is, the high-pressure area corresponds to the real pen (full ink), and the low-pressure area corresponds to the virtual pen (light ink or flying white), providing data support for evaluating the changes in the virtuality and reality of the strokes during the writing process.

[0051] Data transmission unit: It is composed of a Bluetooth unit installed at the end of the pen, and is used to wirelessly transmit the data collected by the above sensors to the data processing module, so as to realize real-time transmission and processing of data and ensure the smoothness of the writing process.

[0052] As the core processing unit of the entire device, the data processing module receives data from the high-speed camera module, inertial measurement unit, pressure sensor module, etc., and performs data fusion and processing.

[0053] The display and feedback module, which is composed of an AR device or a high-resolution screen, is used to project and display the three-dimensional trajectory of the user's writing, the aura heat map and other information generated by the data processing module in real time.

[0054] The above is based on the ideal embodiment of the present invention. Through the above description, relevant personnel can make various changes and modifications without departing from the technical concept of the present invention. The technical scope of the present invention is not limited to the content in the specification, and the technical scope must be determined according to the scope of the claims.

Claims

1. A calligraphy practice method based on an intelligent vision system, characterized in that, Including the steps: S1. Synchronously collect the aerial trajectory, pen movement angle, and pressure distribution data during writing through high-speed vision and an inertial measurement unit; S2. Use spatio-temporal sequence modeling to extract dynamic features from the continuous pen movement trajectory, and quantify the potential energy density and rhythm entropy; S3. Use a lightweight GAN network to generate an ideal virtual pen path, calculate the distance between the user's actual virtual pen and the generated path, and quantify the user's virtual pen deviation; S4. Align the speed-pressure distribution matrices of the user and the master's works using the dynamic time warping algorithm to evaluate the user's rhythm coherence; S5. Project the three-dimensional trajectory and rhythm heat map, and generate rhythm correction and special training suggestions in real time.

2. The calligraphy practice method based on an intelligent vision system according to claim 1, wherein: Track the hydrophobic reflective marker points of the pen tip through a high-speed camera equipped with an 800 - 850nm infrared filter, and synchronously collect the three-dimensional coordinates, speed, acceleration, pen movement angle, and pressure data of the pen tip; among them, the main camera vertically looks down on the paper surface at 40 - 45cm to capture the XY plane trajectory, and the side-view camera is set at an inclination angle of 30 - 60° and 20 - 30cm away from the pen tip to capture the Z-axis suspension height; a nine-axis IMU is set at the center of gravity of the pen shaft to monitor the acceleration and angular velocity, and a micro pressure sensor with a range of 0 - 10N is set on the lower side of the pen shaft grip area to record the pressure distribution.

3. The calligraphy practice method based on an intelligent vision system according to claim 1, characterized in that: The dynamic feature extraction in step S2 includes: Processing spatio-temporal sequences through a Transformer encoder ; Output hidden state sequence , is the motion feature at time t and its relevance to the strokes before and after The quantification indicators include: The potential energy density measures the intensity of dynamic changes. The potential energy density , where is the change amount of the inclination angle per unit time, is the inclination angle change threshold, and T is a certain period of time; Discretize the speed sequence into K intervals; the rhythm entropy is , where is the speed value and is the probability of occurrence.

4. A calligraphy practice method based on an intelligent vision system according to claim 1, characterized in that: In step S3, the generator of the lightweight GAN network uses the current actual stroke end coordinates , and the next actual stroke start coordinates , and outputs an ideal virtual stroke path point sequence by passing through a three-layer fully connected network ; The discriminator inputs the actual virtual stroke path of the user or the generated path , and outputs the probability value that the path conforms to the cursive calligraphy rules; the virtual stroke deviation degree is calculated by using the dynamic time warping algorithm to align the user path with the ideal path , , where is the optimal alignment path, is the Euclidean distance; the deviation degree is quantified , is the total length of the ideal path 5. A calligraphy practice method based on an intelligent vision system according to claim 1, characterized in that: In step S4, the speed sequence extracted from step S2 and the pressure sequence , obtain the standardized speed-pressure matrix of the cursive masters from the cursive calligraphy knowledge base ; Combined speed and pressure , is an adjustment parameter; Calculate the inter-frame Euclidean distance between the user matrix and the master matrix ; i is the user time frame index, , j is the master time frame index, ; Rhythm coherence evaluation: Quantification of rhythm matching degree , the larger the value, the higher the rhythm similarity. is the attenuation coefficient; Similarity score , , is the ratio of the user potential energy density to that of the master, is the adjustment factor.

6. A calligraphy practice method based on an intelligent vision system according to claim 1, characterized in that: In step S5, project the three-dimensional trajectory in real time through an AR device or a screen. The real pen trajectory is displayed in red, the user's virtual pen trajectory is displayed in semi-transparent white, and the ideal virtual pen path is marked in blue.

7. A calligraphy practice device based on an intelligent vision system, based on the calligraphy practice method based on an intelligent vision system according to any one of claims 1-6, characterized in that, Including: A high-speed camera module, including a main camera and a side-view camera; the high-speed camera module is connected to the data processing module; The pen tip marking module, the inertial measurement unit module, the pressure sensing module, and the data transmission unit are all set on the writing brush; the pen tip marking module includes a hydrophobic reflective marker point fixed 3mm from the pen tip; the inertial measurement unit module includes a nine-axis IMU set at the center of gravity of the pen shaft; the pressure sensing module includes a micro pressure sensor with a range of 0 - 10N on the lower side of the pen shaft grip area; The data transmission unit includes a Bluetooth module at the pen tail; The Bluetooth module is connected to the data processing module to transmit the data to the display screen.

8. The calligraphy practice device based on an intelligent vision system according to claim 7, wherein: The main camera and the side-view camera are high-speed cameras above 120fps and are equipped with 800 - 850nm infrared filters; the main camera vertically looks down on the paper surface at 40 - 45cm, and the side-view camera is set at an inclination angle of 30 - 60° and 20 - 30cm away from the pen tip.

9. The calligraphy practice device based on an intelligent vision system according to claim 7, characterized in that: The structure of the hydrophobic reflective marker point from bottom to top includes: a polyimide film base layer with a thickness of 0.05 - 0.07mm, a UV-cured optical glue bonding layer with a thickness of 30 - 50μm, a fluorinated silica nanocoating with a thickness of 20 - 30μm, a contact angle > 150° for blocking ink infiltration, and a ceramic microbead with a diameter of 0.1 - 0.12mm. The thickness of the hydrophobic reflective marker point is less than 0.2mm and the weight is < 0.02g.

10. A calligraphy practice assistance system based on an intelligent vision system, based on the calligraphy practice method based on an intelligent vision system according to any one of claims 1-6, characterized in that, Including: A multi-modal data acquisition module for collecting the three-dimensional aerial trajectory of the pen tip, pen movement angle, pressure distribution, and paper surface contact trajectory data; The dynamic feature extraction module is used to quantify key dynamic indicators and identify real / virtual strokes, pen tip hovering, and virtual connection actions; The virtual stroke path generation module is used to generate an ideal virtual stroke path that conforms to the cursive script rules through a lightweight GAN network; The virtual stroke path evaluation module is used to calculate the deviation between the user's actual virtual stroke and the ideal path; The rhythm coherence evaluation module is used to quantify the rhythm matching degree and rhythm coherence; The real-time feedback module is used to project the three-dimensional trajectory in real time and generate a personalized report.

Images