Automatic calibration method of cutting knife needle, cutting machine, system and storage medium

Abstract

The invention discloses an automatic calibration method and system for a cutting knife needle, a cutting machine and a storage medium. The method comprises the steps that a cutting knife needle is controlled to execute a gradient retraction cutting sequence, and the extension length of the knife needle retracts by a preset gradient length after each time of cutting; collecting a trace image formed on the calibration material by cutting each time through a camera; automatically identifying the shallowest effective cutting trace in the sequence based on image analysis; and calculating the reference zero point of the tool nose according to the sequence number corresponding to the trace, the preset gradient length and the initial extension length. The calibration material has a high-contrast substrate layer and contrast layer structure. The invention further discloses a system for implementing the method, the cutting machine comprising corresponding hardware and a storage medium. According to the invention, full automation of the calibration process is realized, dependence on artificial experience is eliminated, and the method has the advantages of high calibration precision, high efficiency, good consistency and the like.

Description

Technical Field

[0001] This invention relates to the field of cutting equipment and automated control technology, and more specifically, to an automatic calibration method for a cutting needle, a cutting machine and system for performing the method, and related computer storage media. Background Technology

[0002] In modern precision machining fields, such as die-cutting, flexible circuit board processing, label making, and composite material layer cutting, CNC cutting machines have become core production equipment. As the terminal component that directly performs the cutting function, the precise control of the cutting needle's extension length is crucial to ensuring consistent cutting depth, edge quality, and preventing material damage. However, in actual production, due to mechanical wear of the cutting needle, replacement and installation errors, and the diversity of the types, thicknesses, and layered structures of the materials to be processed, the "zero point" of the cutting needle (i.e., the reference position where the cutting tip begins to effectively contact and cut into the material) must be repeatedly and precisely calibrated.

[0003] Traditional calibration methods heavily rely on the skills and experience of operators. Typically, technicians manually adjust the cutting needle position through trial cuts, visually observing the depth and integrity of the cuts to determine if the optimal cutting point has been reached. This method results in highly subjective calibration outcomes, significantly influenced by operator vision, experience, and ambient lighting. This leads to inconsistencies in equipment condition after calibration by different batches or operators, resulting in significant fluctuations in product accuracy. Furthermore, the calibration process is time-consuming and labor-intensive, especially when frequently changing cutting needles or materials, severely limiting equipment utilization and production efficiency.

[0004] Therefore, there is an urgent need in this field for a solution that can achieve fully automatic zero-point calibration of the cutting needle. Summary of the Invention

[0005] The purpose of this invention is to provide an automatic calibration method, cutting machine, system, and storage medium for cutting needles, which realizes full automation and objectification of the calibration process, eliminates the reliance on human observation and manual experience, and thus greatly improves calibration accuracy, efficiency, and consistency.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, an automatic calibration method for a cutting needle is provided, comprising: controlling the cutting needle to move a preset length from a reference origin along the extension direction, and then performing a first cut on a calibration material; controlling the cutting needle to retract according to a preset gradient length based on the extension length of the previous cut and then performing the next cut, forming a gradient retraction cutting sequence; acquiring images of the cutting marks formed on the calibration material after each cut, and automatically identifying the shallowest effective cutting mark that meets preset conditions from the gradient retraction cutting sequence based on image analysis; and determining the blade tip reference zero point of the cutting needle according to the sequence number of the gradient retraction cutting sequence corresponding to the shallowest effective cutting mark, the preset gradient length, and the preset length.

[0007] Preferably, the calibration material includes a substrate layer and a contrast layer disposed on its surface, the contrast layer and the substrate layer having a preset contrast in optical characteristics, and the cutting mark being the portion of the substrate layer exposed after the contrast layer is cut.

[0008] Preferably, the substrate layer is white, and the contrast layer is a black coating or printed layer.

[0009] Preferably, the step of automatically identifying the shallowest effective cutting mark based on image analysis includes: processing the image to extract the cutting mark region; calculating at least one image feature parameter of the cutting mark region; and determining the cutting mark that meets the preset conditions based on the change of the image feature parameter in the gradient shrinkage cutting sequence.

[0010] Preferably, the image processing includes the following steps: performing distortion correction, illumination equalization, and noise reduction on the acquired image; performing binarization segmentation on the processed image to obtain a binary image of the cutting marks; performing morphological dilation and closure operations on the binary image to connect the broken cutting marks; and searching and filtering out the contours of the cutting marks from the processed image.

[0011] Preferably, the image feature parameters include at least one of the following: the average width of the cutting mark region, the contrast difference between the cutting mark region and the background region, and the edge continuity evaluation value of the cutting mark region.

[0012] Preferably, the preset gradient length is dynamically adjusted according to the changes in the cutting trace features obtained from the image analysis; the dynamic adjustment includes: reducing the value of the preset gradient length when the contrast of the cutting trace is detected to begin to decrease.

[0013] Preferably, it further includes: acquiring images along the cutting path in real time during the cutting task; The extension length of the cutting needle is adjusted in real time based on image analysis results.

[0014] Preferably, the method further includes: recording the zero-point reference data of the cutting tip determined by each calibration; establishing a wear prediction model of the cutting needle based on the data; and outputting maintenance prompts based on the wear prediction model.

[0015] Preferably, the image acquisition is accomplished by a camera mounted on the cutting machine, and the method further includes: dynamically controlling the brightness of the auxiliary light source according to the ambient lighting conditions to optimize the image acquisition quality.

[0016] In a second aspect, a cutting machine is also provided, comprising: a cutting actuator including a blade holder, a drive unit, and a cutting needle that is driven by the drive unit to extend and retract relative to the blade holder; an image acquisition device configured to acquire an image of the area of ​​action of the cutting needle; and a control processing unit communicatively connected to the drive unit and the image acquisition device, wherein the control processing unit is configured to perform the automatic calibration method as described in the first aspect.

[0017] Preferably, it also includes an auxiliary light source, the light emission direction of which is directed toward the framing area of ​​the image acquisition device.

[0018] Preferably, the image acquisition device is an industrial camera or a module integrated with a vision sensor.

[0019] Thirdly, the present invention also provides an automatic calibration system for a cutting needle, comprising: a cutting control module for controlling the cutting needle to execute a gradient retraction cutting sequence; a visual recognition module for acquiring and analyzing trace images generated by each cut to automatically identify the shallowest effective cutting trace; a zero-point calculation module for calculating the blade tip reference zero point based on the recognition result; and an adaptive control module for dynamically adjusting the execution parameters of the gradient retraction cutting sequence according to the analysis result of the visual recognition module.

[0020] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0021] This invention utilizes a gradient-retraction mechanical test sequence and machine vision-based automatic recognition and judgment to automate the entire calibration process. The calibration material provides a stable and optimized recognition environment, while image processing algorithms analyze cutting marks and make automatic decisions based on objective data. This eliminates reliance on manual skills and significantly improves calibration accuracy, efficiency, and consistency. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below.

[0023] Figure 1 This is a flowchart illustrating the automatic calibration method for the cutting needle in the embodiment.

[0024] Figure 2 This is a schematic diagram of the half-section structure of the cutting actuator in the embodiment.

[0025] Figure 3 This is a schematic diagram of the calibration material with serial numbers forming a gradient shrinkage cutting sequence in the embodiment. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the specific implementation methods of this invention will be comprehensively and deeply described below in conjunction with the control logic of this invention and the hardware system on which this logic is implemented. This section will describe in detail how the control unit, through software programs and algorithms, coordinates and schedules various sensors, actuators, and mechanical mechanisms to complete the operation sequence of the automatic calibration method for the full-cutting needle. The aim is to enable those skilled in the art to fully understand and reproduce this invention.

[0027] Example Please see Figure 1 , Figure 1 This is a flowchart illustrating the automatic calibration method for a cutting needle provided in this application embodiment. This method can be applied to a cutting machine that can cut sheet materials such as films, fabrics, and paper. The automatic calibration method for the cutting needle may include: In S1, after controlling the cutting needle to move a preset length from the reference origin along the extension direction, the first cut is performed on the calibration material.

[0028] It is understood that the reference origin is a known, repeatable mechanical or electrical reference position of the cutting needle, such as the position of the internal limit switch triggered when the needle is fully retracted, the end position of the stroke, or a zero position set by the encoder of the drive motor.

[0029] The cutting needle is controlled to move a preset length L along the extension direction from the reference origin. This preset length L must be set to ensure that the tip of the cutting needle, at this position, can completely penetrate the surface coating of the calibration material and cut into its substrate to a certain depth, thereby forming a clear cutting mark. For example, the preset length L can be set to 0.70 mm.

[0030] Subsequently, the cutting needle is controlled to extend to this length to perform the first cut on the calibration material, forming an initial test cut mark. The test cut mark can be a simple geometric shape, such as a short line, a circle, or a cross.

[0031] In S2, please refer to Figure 3The cutting needle is controlled to retract according to a preset gradient length based on the extension length of the previous cut and then perform the next cut, forming a gradient retraction cutting sequence.

[0032] Understandably, after the first cut, the cutting needle is controlled to retract by a preset gradient length f, where f is much smaller than L, based on its current extension length. For example, the preset gradient length f can be set to 0.02mm, 0.01mm, or 0.005mm. After the retraction operation is completed, the cutting needle is controlled to perform a second cut with this new, shorter extension length (i.e., Lf), forming a second test cut mark.

[0033] Subsequently, the "retraction-cutting" process is repeated: after each cut, the cutting needle retracts by the same preset gradient length f based on the previous extension length, and then performs the next cut with the new retracted length. This series of operations constitutes a gradient retraction cutting sequence, in which the needle extension length Si corresponding to the i-th cut (i=1, 2, 3...) can be expressed as: Si = L – (i-1) × f. As i increases, Si decreases, and the cutting depth becomes shallower accordingly.

[0034] In S3, images of the cutting marks formed on the calibration material after each cut are acquired; based on image analysis, the shallowest effective cutting marks that meet preset conditions are automatically identified from the gradient retraction cutting sequence.

[0035] In some embodiments, the worktable of the cutting machine is provided with a dedicated cutting area and a dedicated vision inspection area. After the calibration material is cut, the Y-axis linear motor module integrated on the cutting machine starts working, transferring the calibration material with cutting marks to the dedicated vision inspection area via a clamp or roller. This area is equipped with a background light source (such as coaxial light or backlight) to ensure a consistent imaging environment. An industrial camera fixed above the dedicated vision inspection area triggers shooting after the calibration material is in place, acquiring a raw digital image containing the complete white line.

[0036] In this embodiment, the calibration material has a high-contrast structure consisting of a white substrate and a black surface coating. After each cut, an industrial camera, fixedly mounted above the cutting needle or near the blade holder, automatically acquires an image containing the cutting mark under illumination from a matching light source. To obtain stable images, the acquisition action is triggered immediately after the needle is lifted, and the brightness of the light source can be adaptively adjusted according to the ambient light.

[0037] The image analysis-based automatic recognition process includes the following sub-steps: For each acquired image, a standardized processing procedure is performed to quantify the visual features of the cutting marks. First, the color image is converted to a grayscale image and Gaussian filtering is applied. Next, based on a preset test pattern template, the region of interest (ROI) containing the cutting marks is located in the image. For this ROI, an adaptive thresholding algorithm (e.g., Otsu's method) is used for binarization segmentation to separate the cutting marks (the exposed part of the white substrate) from the background (black coating). Then, an edge detection algorithm (e.g., the Canny operator) is used to extract the precise contour of the cutting marks in the binary image.

[0038] Based on the extracted contours, one or more image feature parameters are calculated to objectively assess the significance of the cutting marks. These mainly include: Contrast Difference (C): the difference between the average gray value of pixels within the contour region and the average gray value of pixels in the adjacent background region outside the contour. This value directly reflects the visual contrast intensity between the "white mark" and the "black background".

[0039] Physical width of the trace (W): Select multiple points along the cutting trajectory, calculate the width of the contour in the normal direction at that point, take the average value, and convert it into the actual physical size (e.g., millimeters) in combination with the camera calibration parameters.

[0040] Contour continuity index (E): assesses whether the extracted contour is complete, continuous, and without breaks. For example, it can be quantified by calculating the ratio of the actual perimeter of the contour to its theoretical perimeter after fitting it to an ideal geometric shape (such as a straight line or circle). The closer the ratio is to 1, the better the continuity.

[0041] After each cut, the system records the corresponding cut sequence number i and the calculated set of feature parameters (Ci, Wi, Ei).

[0042] The system does not judge the trace of a single cut in isolation, but analyzes the changes of the above feature parameters with the increase of the cut number i (i.e., the extension length decreases) throughout the entire gradient shrinkage cut sequence. Generally, as the cut depth becomes shallower, the contrast difference Ci and the trace width Wi will show a monotonically decreasing trend.

[0043] In some embodiments, an image processing process based on OpenCV can also be used for feature extraction. The line contours with sub-pixel accuracy are robustly extracted from the complex background. By using the pre-calibrated camera internal parameters and distortion coefficients, the cv2.undistort function is called to perform geometric correction on the original image, eliminating the lens barrel distortion and ensuring the geometric authenticity of the measurement reference. cv2.GaussianBlur or median filtering is used to reduce random noise. Adaptive histogram equalization or cv2.normalize is used to balance the uneven illumination of the entire field of view, weakening the interference of reflection and shadow. Since the contrast between the target (white line) and the background (black paper) is extremely high, automatic thresholding can be used for binarization to convert the image into pure black (background, pixel value 0) and pure white (line, pixel value 255). Using cv2.dilate, the white area is dilated with a small structuring element (such as a 3x3 rectangle), which can bridge the small fractures caused by incomplete coating peeling on the line. The closing operation (i.e., dilating first and then eroding) can smooth the line edges, fill small holes, and connect adjacent fractured parts, finally obtaining a continuous, full, and smooth-edged white line area, greatly improving the stability of subsequent contour finding. The cv2.findContours function is called to find all external contours in the binary image. The target line contours are filtered out by area, aspect ratio, etc. The number of the filtered line contours is the actual cutting number.

[0044] In this embodiment, the "shallowest effective cutting mark that meets the preset conditions" is automatically determined by the following logic: The system sets an effective threshold T for the key feature parameter (preferably the contrast difference C). For example, T = 30 can be set (within the gray scale range of 0 - 255). The system then analyzes the sequence data to find the cutting sequence number N that meets the following conditions: CN ≥ T and C(N + 1) < T. That is, in the gradient retraction cutting sequence, the contrast of the mark generated by the Nth cut is still not lower than the effective threshold, while the contrast of the mark generated by the (N + 1)th cut immediately following it is lower than this threshold. At this time, the system determines the mark corresponding to the Nth cut as the "shallowest effective cutting mark".

[0045] Its purpose is to identify the critical state where the tip protrusion length just changes from "being able to effectively cut through the black coating to form a visible white mark" to "not being able to effectively cut through the coating or the mark cannot be reliably identified". The protrusion length SN corresponding to the Nth cut = L - (N - 1) f, which is this critical length.

[0046] In some implementations, multi-feature fusion decision-making can be used to improve the robustness of the determination. For example, the mark of the i-th cut is considered "valid" only when all three conditions are met simultaneously: Ci ≥ T1, Wi ≥ T2, and Ei ≥ T3. Similarly, the shallowest valid cut mark is determined by finding the last cut number N that satisfies all the conditions.

[0047] In other implementations, the system can also determine that it has entered the critical region when it detects a sudden increase in the rate of decrease of a feature parameter (such as contrast difference), and can dynamically switch the preset gradient length f to a smaller value to perform a finer scan, thereby more accurately locating the critical point N.

[0048] Through the above automated image acquisition, processing, and logical judgment process, the system is able to identify the shallowest effective cutting trace in the gradient shrinkage cutting sequence without relying on manual observation and subjective experience.

[0049] In S4, the blade tip reference zero point of the cutting needle is determined based on the sequence number of the gradient retraction cutting sequence corresponding to the shallowest effective cutting mark, the preset gradient length, and the preset length.

[0050] In this embodiment, the shallowest effective cutting mark automatically identified in step S3 corresponds to the Nth cut in the gradient retraction cutting sequence. When this cut occurs, the extension length SN of the cutting needle is: SN = L - (N-1) f, where L is the preset length and f is the preset gradient length.

[0051] The blade tip reference zero point is defined as the reference position where the tip of the cutting needle just contacts the surface of the material to be cut and is about to begin effective cutting. From a physical perspective, the extension length SN of the Nth cut identified in step S3 is the minimum length by which the blade tip can form a visible and effective cutting mark on the calibration material. Therefore, this extension length SN is numerically equal to the distance from the reference origin to the material surface (i.e., the blade tip reference zero point).

[0052] Therefore, the position P of the tool tip reference zero point relative to the reference origin can be determined by the following formula: P = SN = L - (N-1) f. Example. Assume the preset length L is set to 0.70 mm and the preset gradient length f is set to 0.02 mm. During automatic recognition, the system determines the mark from the 15th cut (N=15) as the shallowest effective cut mark. Substitute the values ​​into the formula to calculate: P = 0.70 mm – (15-1) × 0.02 mm = 0.70 mm – 14 × 0.02 mm = 0.70 mm – 0.28 mm = 0.42 mm. The calculation result indicates that moving 0.42 mm along the extension direction from the reference origin (such as the mechanical zero point) is the tool tip reference zero point determined for this calibration.

[0053] The control processing unit stores the calculated zero-point position P (e.g., 0.42mm in the example above) as a parameter in the system's memory. During subsequent actual cutting tasks, for any material to be cut, the system will retrieve the pre-stored target cutting length D corresponding to that material. When controlling the movement of the cutting needle, the system will use the zero-point reference P obtained from this calibration as the absolute starting point, controlling the cutting needle to extend an additional length D, thereby ensuring the cutting tip reaches the precise predetermined cutting depth. This process achieves "one-time calibration, shared zero point," ensuring consistent accuracy for different materials and cutting depth requirements.

[0054] In some implementations, if minor differences in system structure or algorithm are considered (e.g., the initial state of the first cut is regarded as having completed one retraction), the calculation formula can be expressed as P = L – N × f. Its physical essence is consistent with the aforementioned method, which determines the zero-point offset based on the critical number of cuts, the initial extension amount, and the retraction step size. Regardless of the equivalent mathematical expression used, the core principle is to obtain the precise zero-point position through calculation based on the automatically identified critical sequence number.

[0055] Through the above steps, this embodiment completes a fully closed-loop automated process from automatically executing test sequences and visually recognizing critical traces to accurately calculating the zero-point position, replacing the traditional method that relies on manual visual inspection and adjustment, and achieving highly efficient zero-point calibration of the tool tip.

[0056] The calibration material includes a substrate layer and a contrast layer disposed on its surface. The contrast layer and the substrate layer have a preset contrast in optical characteristics. The cutting mark is the portion of the substrate layer exposed after the contrast layer is cut.

[0057] It is understood that the calibration material is a consumable component. The substrate layer constitutes the carrier of the calibration material and the final imaging portion of the cutting marks. The substrate layer is preferably made of a material with high reflectivity, pure color, and homogeneous surface. For example, the substrate layer is white, with a uniform, easily peelable black coating (such as toner or a special coating) covering the surface as a contrast layer. When the tool cuts through, the white of the substrate layer is clearly exposed, forming a high-contrast white line.

[0058] The contrast layer is tightly adhered to the surface of the substrate layer, and its characteristic is that it forms a stable optical contrast with the substrate layer. The contrast layer is typically dark, preferably black. In a specific example, the contrast layer is a black ink layer formed by screen printing or precision coating. The contrast layer needs to be thin enough to be easily scratched by a knife tip, while also being sufficiently uniform and completely covering the underlying white layer to prevent it from showing through.

[0059] The system first controls the tool holder to move to a fixed position. When the cutting needle performs cutting, the pressure of the blade tip cuts or peels away the local contrast layer, exposing the white substrate layer underneath. This creates a clear "white line" or "white mark" on a black background, which is the cutting mark. This "black background with white mark" structural design generates pixel grayscale difference (i.e., contrast difference) in the image, enabling image processing algorithms to detect, segment, and quantify the mark very stably and easily, overcoming the recognition interference problems caused by uneven color, texture, and reflectivity of traditional random materials. Therefore, this calibration material provides a standardized and optimized recognition scenario for machine vision systems at a low cost, forming the physical basis for automatically determining the "shallowest effective cutting mark."

[0060] It is understandable that the phrase "determining cutting marks that meet preset conditions based on the changes in image feature parameters in the gradient shrinking cutting sequence" means that the system does not view the image features of a single cut in isolation, but rather plots the same feature parameter values ​​(such as C1, C2, C3, ... Ci) calculated for each cut in the entire sequence as a change curve for analysis. As the cutting depth gradually decreases (corresponding to an increase in the sequence number i), parameters C and W are expected to show a monotonically decreasing trend. The preset condition essentially defines a critical state between "valid" and "invalid". For example, the system sets the effective threshold T for contrast difference C to 35. The judgment logic is: scan the sequence from back to front, and find the last cutting sequence number N that satisfies Ci ≥ T. This means that the mark of the Nth cut is still considered clear and valid, while the mark of the (N+1)th cut is judged as invalid or unreliable because the contrast is below the threshold. Thus, the Nth cut is automatically identified as the "shallowest valid cutting mark". This automated judgment method, based on sequence trends and threshold comparison, eliminates the interference of random noise in a single image and mimics the logic of the human eye finding the critical point through multiple attempts, but the results are more accurate and consistent.

[0061] In some embodiments, the preset gradient length is dynamically adjusted based on the changes in the cutting trace features obtained from the image analysis; the dynamic adjustment includes reducing the value of the preset gradient length when the contrast of the cutting trace begins to decrease. Here, the preset gradient length f is a fixed value. However, in this preferred embodiment, f can be dynamically adjusted. Specifically, in step S3, the system monitors the rate of change of image feature parameters (especially the contrast difference C) in real time. For example, a rate of change threshold R is set. When the system detects that the rate of decrease in the C value exceeds R for several consecutive cuts (i.e., (C_i - C_{i+1}) / C_i>R), it determines that it has entered the "critical change zone," meaning the blade tip extension length is close to the critical point where it can just break through the coating. At this time, the system automatically reduces the current preset gradient length f (e.g., 0.02 mm) to a finer value f' (e.g., 0.005 mm). Subsequent cutting tests will continue with a retraction scan using f' as the step size. The benefits of this approach are: in the early stages, a larger step size is used to quickly approach the critical region, saving time; in the later stages, a smaller step size is used to precisely determine the critical point, improving the zero-point positioning accuracy, thus achieving a balance between speed and accuracy.

[0062] In some implementations, the method further includes: acquiring images along the cutting path in real time during the cutting task; and adjusting the extension length of the cutting needle in real time based on the results. It is understood that during the actual cutting operation, a camera (or another dedicated inspection camera) can take real-time or sampled images along the cutting path to obtain edge images of the cut area. The image processing module analyzes the quality characteristics of the cutting edge in real time, such as the degree of edge burrs and the height of material lifting (edge ​​curling). If the analysis finds that the edge quality parameters (such as burr width) exceed the preset normal range, the control unit can immediately fine-tune the extension length of the cutting needle for compensation. For example, if excessive burrs are detected, it means the cut is too shallow, and the system can immediately increase the needle extension length by a small compensation amount (such as 0.005mm) and apply this compensation in subsequent cuts. This forms a closed-loop process control based on visual feedback, which can effectively cope with dynamic interferences such as slight changes in material thickness and fluctuations in platform flatness.

[0063] In some implementations, it also includes: recording the tool tip reference zero point data determined by each calibration; Based on the data, a wear prediction model for the cutting needle is established; maintenance prompts are output according to the wear prediction model. This approach enables the system to perform predictive maintenance. The control unit records information such as the zero-point reference position P (e.g., 0.42mm) obtained after each successful calibration, the total number of cuts during calibration, and the needle ID in a historical database. By analyzing the P value over a long time series, it can be observed that the P value typically increases gradually with needle use (because the cutting tip wears, requiring a longer extension distance to achieve the same cutting effect). The system can use a simple linear regression time series model to fit a wear curve of "zero-point offset - usage intensity (e.g., total cutting mileage or number of cuts)". When the zero-point offset predicted by the model is about to exceed the maximum allowable tolerance of the equipment, or when the remaining life calculated based on the curve slope is lower than the safety threshold, the system will proactively display a maintenance prompt on the human-machine interface, such as "Needle wear is approaching its limit; it is recommended to check or replace after N cuts." This transforms passive maintenance into proactive prediction, reducing unplanned downtime.

[0064] In some implementations, the image acquisition is accomplished using a camera mounted on the cutting machine. The method further includes dynamically controlling the brightness of an auxiliary light source based on ambient lighting conditions to optimize the image acquisition quality. This approach ensures consistent image acquisition quality under any lighting conditions. Specifically, an ambient light sensor can be integrated into or near the camera. Before each image acquisition is triggered, the system reads the ambient light intensity value I_env. The control unit stores a mapping table or calculation formula for the brightness value I_led of the auxiliary light source (such as a ring LED) required to obtain optimal image contrast under different ambient light intensities. The system looks up the table or calculates based on I_env in real time and automatically adjusts the drive current of the auxiliary light source to stabilize its output brightness I_led at the set value. For example, the brightness of the supplementary light is automatically reduced in bright environments and automatically increased in dim environments, thereby ensuring that the acquired cutting mark image has stable and optimal exposure and contrast, avoiding overexposure, underexposure, or insufficient contrast due to changes in lighting, which could affect the reliability of feature extraction and judgment. This improves the robustness of the system in different working environments.

[0065] This embodiment also provides a cutting machine, see reference. Figure 2 As shown, Figure 2This is a schematic diagram of a half-section of the cutting actuator. The cutting machine includes: a cutting actuator comprising a tool holder, a drive unit, and a cutting needle that extends and retracts relative to the tool holder, driven by the drive unit; an image acquisition device configured to acquire images of the area of ​​action of the cutting needle; and a control processing unit communicatively connected to the drive unit and the image acquisition device, configured to execute the automatic calibration method as described in the first aspect. For example, the control processing unit may include an Android main control board and an ST32 control board, wherein the Android main control board serves as the core computing and scheduling center of the system. It is responsible for controlling the operation of the vision acquisition module, running advanced image processing algorithms based on OpenCV, controlling the logical sequence of the entire calibration process, providing a human-machine interface, and communicating data with the lower-level CNC system. The ST32 control board serves as the underlying real-time control unit. It receives instructions from the Android board and is responsible for controlling all actions of various mechanisms that have high timing and real-time requirements.

[0066] For example, the image acquisition device can be an ultra-wide-angle, high-resolution (e.g., 2 megapixels or higher) CMOS sensor camera to ensure that distortion-free, high-definition line images can be captured on paper. An auxiliary light source is required to create a stable and uniform lighting environment. In some embodiments, the light emission direction of the auxiliary light source is directed towards the viewing area of ​​the image acquisition device.

[0067] It should be further noted that the cutting machine provided in this embodiment is a product entity that fully integrates the aforementioned automatic calibration method and system at the physical device level. This cutting machine not only includes the conventional mechanical and electrical components required to perform the cutting task, but also integrates the dedicated hardware and software necessary to achieve automatic calibration.

[0068] Specifically, the cutting actuator is the terminal unit that completes the cutting action. The tool holder is a mechanical structure used to fix and guide the cutting needle, and its bottom end typically has a reference plane. The drive unit is usually a stepper motor or a closed-loop servo motor, which, in conjunction with a ball screw or linear motor transmission mechanism, controls the extension and retraction of the cutting needle relative to the tool holder. The cutting needle is a replaceable consumable part, and its tip obtains positioning capability through this drive unit.

[0069] The image acquisition device includes an industrial camera fixedly mounted on the side of the tool holder or at a suitable position on the frame, with its optical axis calibrated and vertically aligned with the calibration area below the tool tip. To obtain clear images of the cutting marks, the device also integrates an auxiliary light source, preferably a ring-shaped LED light surrounding the lens, to provide shadowless and uniform illumination to the shooting area. The image acquisition device is connected to the control processing unit via a high-speed interface (such as USB 3.0 or GigE).

[0070] The control processing unit can be a high-performance industrial computer, an embedded industrial control computer, or a dedicated controller based on FPGA / DSP. In hardware, this unit controls the drive unit via a drive interface card and connects to the image acquisition device via a vision interface card. At the software level, it internally stores and runs a control program containing all the logic of the automatic calibration method. This control processing unit can be programmed to: send instructions to the drive unit, directing the cutting needle to execute a preset "gradient retraction cutting sequence"; trigger the image acquisition device to synchronously take pictures after each cut and receive the returned image data; call the embedded image processing algorithm to analyze the image, calculate feature parameters, and automatically determine the shallowest effective cutting mark; calculate a new tool tip reference zero point based on the determination result and update the system parameters; and control all cutting depths based on this zero point in subsequent processing tasks.

[0071] Furthermore, this control processing unit can be expanded to support advanced functions such as dynamic gradient adjustment (changing the retracement step size in real time based on image analysis), ambient light adaptation (adjusting the brightness of the auxiliary light source), and scalpel health management (recording zero-point history and predicting lifespan). Users can initiate calibration, set parameters, and view results through a human-machine interface (such as a touchscreen) connected to this control processing unit.

[0072] This embodiment also provides an automatic calibration system for cutting needles, including: a cutting control module for controlling the cutting needle to execute a gradient retraction cutting sequence; a visual recognition module for acquiring and analyzing trace images generated by each cut to automatically identify the shallowest effective cutting trace; a zero-point calculation module for calculating the blade tip reference zero point based on the recognition result; and an adaptive control module for dynamically adjusting the execution parameters of the gradient retraction cutting sequence according to the analysis result of the visual recognition module.

[0073] It should be further noted that the automatic calibration system for cutting needles provided in this embodiment is a modular implementation of the logical steps and functions described in the aforementioned method embodiments at both the hardware and software levels. This system can serve as the core of an independent calibration device or be integrated as a functional unit into an existing intelligent cutting machine control system.

[0074] Specifically, the cutting control module is primarily responsible for interacting with the motion hardware of the cutting machine. Physically, it can correspond to the existing CNC system of the cutting machine or a newly added dedicated motion controller. Logically, it includes a drive unit interface for controlling the extension and retraction of the cutting needle, an axis control interface for controlling the movement of the platform, and program logic for executing the gradient retraction cutting sequence. This module receives sequence parameters (such as the initial preset length L, the dynamically adjusted gradient length f, etc.) from the adaptive control module and accurately converts them into pulse commands for the drive motor, controlling the cutting needle to complete a series of "extension-cutting-retraction" mechanical actions.

[0075] The hardware foundation of the visual recognition module is the aforementioned industrial camera, lens, light source, and control circuitry. Its software runs on the system's image processor or main controller, encapsulating a complete image processing pipeline algorithm, including functional units such as image acquisition driving, preprocessing, ROI localization, binarization segmentation, contour extraction, and feature parameter calculation. The module's responsibility is to output one or more quantized feature parameter values ​​(such as Ci, Wi, Ei) for each cut, and ultimately, based on the variation patterns of these parameters in the sequence, execute preset judgment logic (such as threshold comparison) to output a cut sequence number N representing the "shallowest effective cut mark."

[0076] The zero-point calculation module is a data processing and calculation unit, typically implemented as a software algorithm. It receives the output (i.e., sequence number N) from the visual recognition module and, combined with the current sequence parameters (L, f) provided by the cutting control module, performs a simple arithmetic operation (P = L - (N-1)). f), calculates the precise zero-point reference position P of the cutting tip. This module is also responsible for securely storing this zero-point data in the system's non-volatile memory area for subsequent cutting tasks to access.

[0077] The adaptive control module monitors the entire process and dynamically optimizes it based on real-time feedback. This module receives intermediate data from the visual recognition module in real time, such as the feature parameter values ​​and their trends after each cut. Its embedded decision logic (e.g., based on the rate of change) determines whether to adjust subsequent test parameters. For example, if it detects that the contrast difference is decreasing rapidly, it immediately sends a command to the cutting control module to change the preset gradient length f from 0.02mm to 0.005mm. Furthermore, this module may also manage calibration trigger conditions, the recording and analysis of needle wear data, and interaction with the user interface (e.g., displaying calibration progress and alarm information).

[0078] These four modules are closely connected and work together via an internal system bus or communication interface (such as EtherCAT, CAN, or an internal data sharing area). Their typical workflow is as follows: the adaptive control module initializes parameters and triggers the process → the cutting control module performs the first cut → the vision recognition module acquires and processes the image, feeding back the feature parameters to the adaptive control module → the adaptive control module analyzes the image and determines the next set of parameters → the cutting control module performs the next cut... This cycle continues until the vision recognition module identifies the shallowest effective cutting mark and sends a command to the zero-point calculation module → the zero-point calculation module completes the calculation and stores the result → the adaptive control module reports calibration completion to the user.

[0079] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in the first aspect. It should be further noted that the computer-readable storage medium provided in this embodiment is the physical carrier and implementation method of the automated calibration method at the software level. It solidifies the intelligent steps and logic described in the foregoing method embodiments, executed by the control processing unit, into executable code, enabling any cutting device or general-purpose computing device equipped with compatible hardware (including the driving unit and image acquisition device) to obtain and implement the automatic calibration function after loading and running the program.

[0080] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An automatic calibration method for a cutting needle, characterized in that, include: After controlling the cutting needle to move a preset length from the reference origin along the extension direction, the first cut is performed on the calibration material; The cutting needle is controlled to retract according to a preset gradient length based on the extension length of the previous cut and then perform the next cut, forming a gradient retraction cutting sequence; Images of the cutting marks formed on the calibration material after each cut are acquired. Based on image analysis, the shallowest effective cutting marks that meet the preset conditions are automatically identified from the gradient retraction cutting sequence. The zero point of the cutting needle tip is determined based on the sequence number of the gradient retraction cutting sequence corresponding to the shallowest effective cutting mark, the preset gradient length, and the preset length.

2. The method according to claim 1, characterized in that, The calibration material includes a substrate layer and a contrast layer disposed on its surface. The contrast layer and the substrate layer have a preset contrast in optical characteristics. The cutting mark is the portion of the substrate layer exposed after the contrast layer is cut.

3. The method according to claim 1, characterized in that, The image analysis-based automatic identification of the shallowest effective cutting marks includes: The image is processed to extract the cutting mark area; Calculate at least one image feature parameter of the cutting mark region; Based on the changes in the image feature parameters in the gradient shrinkage cutting sequence, the cutting marks that meet the preset conditions are determined.

4. The method according to claim 3, characterized in that, The image feature parameters include at least one of the following: the average width of the cutting mark region, the contrast difference between the cutting mark region and the background region, and the edge continuity evaluation value of the cutting mark region.

5. The method according to claim 1, characterized in that, The preset gradient length is dynamically adjusted based on the changes in the cutting trace features obtained from the image analysis; the dynamic adjustment includes: reducing the value of the preset gradient length when the contrast of the cutting trace is detected to begin to decrease.

6. The method according to claim 1, characterized in that, Also includes: During the cutting task, images along the cutting path are acquired in real time; The extension length of the cutting needle is adjusted in real time based on image analysis results.

7. The method according to claim 1, characterized in that, Also includes: Record the zero-point reference data of the tool tip determined in each calibration; A wear prediction model for the cutting needle is established based on the data. The wear prediction model outputs maintenance tips.

8. A cutting machine, characterized in that, include: A cutting actuator includes a tool holder, a drive unit, and a cutting needle that is driven by the drive unit to extend and retract relative to the tool holder; An image acquisition device is configured to acquire images of the area of ​​action of the cutting needle; A control processing unit is communicatively connected to the drive unit and the image acquisition device, and the control processing unit is configured to perform the automatic calibration method as described in any one of claims 1 to 7.

9. An automatic calibration system for cutting needles, characterized in that, include: The cutting control module is used to control the cutting needle to execute a gradient retraction cutting sequence; The visual recognition module is used to acquire and analyze the trace images generated by each cut in order to automatically identify the shallowest effective cut trace; The zero-point calculation module is used to calculate the blade tip reference zero point based on the recognition results; An adaptive control module is used to dynamically adjust the execution parameters of the gradient shrinking and cutting sequence based on the analysis results of the visual recognition module.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images