A method and system for online detection of laser perovskite cell etching quality

Abstract

The present invention discloses an online detection method and system for the etching quality of laser perovskite cells. The method adopts a dual-camera alternating working mode, installs high-resolution CCD camera assemblies on both sides of the laser processing head in the X direction, eliminates motion direction occlusion, combines the submicron optical imaging of the CCD camera assembly with an image processing algorithm to analyze the etching quality, proposes local adaptive threshold segmentation, uses the Otsu method to adaptively calculate the binarization threshold, and then uses a morphological opening operation to remove irrelevant information. Then, an edge-finding logic algorithm is used to find the two edges of the etching line, and finally, the two edges and a center line are fitted. The three lines are used to analyze the etching quality. In addition, the etching quality is detected to perform real-time power adjustment, position calibration, and defect location marking. The high-resolution optical detection system has a fast calculation speed and can perform detection while processing, realizing a "processing-detection-correction" closed-loop control.

Description

Technical Field

[0001] The present invention relates to the technical field of laser line engraving, and in particular to an online detection method and system for the etching quality of a laser perovskite cell. Background Art

[0002] Perovskite solar cells, a novel photovoltaic technology based on perovskite structural materials, have become a research hotspot in the energy sector in recent years due to their high efficiency, low-cost potential, and flexible application scenarios. In the manufacture of perovskite solar cells, laser scribing techniques (P1, P2, and P3) are key processes in the modular tandem structure. The P1 scribing separates the bottom transparent conductive oxide (TCO, such as ITO or FTO) layer to form independent sub-cell regions. This separates the continuous TCO layer into multiple units, laying the foundation for subsequent tandem structures. The technical challenge lies in the precise control of laser energy to ensure complete severing of the TCO layer while avoiding damage to the substrate material (such as glass).

[0003] To improve processing efficiency, existing laser etching equipment generally utilizes multi-path parallel high-speed scribing technology. However, existing methods for inspecting P1 scribing results suffer from issues such as defect detection lag and missed micron-level defects. Traditional scanning electron microscope (SEM) or electrical probe inspection requires destructive sampling and takes hours. This lack of real-time feedback on etch residues (such as incomplete electrode or perovskite layers) leads to the risk of batch rework. The functional layers of perovskite cells are only a few hundred nanometers thick, and incomplete etching can produce submicron-level residues (such as metal debris <1μm) that exceed the resolution limits of conventional optical inspection, potentially posing a risk of internal short circuits in the cells.

[0004] Patent document CN217739435U discloses a device for automatically detecting the P1 marking effect of perovskite battery modules. The test sequence of the detection probes is that two rows of detection probes move horizontally along the length direction of the fixed frame, respectively testing the resistance of the areas on both sides of the even and odd PI marking lines. This is simple, convenient and effective, and can automatically detect the resistance of the areas on both sides of each P1 marking line. It automatically determines whether the front electrode is broken based on the resistance value. It is suitable for detecting the P1 marking condition of any size perovskite battery module formed by small battery modules in series or parallel, filling the gap in the detection of the P1 marking condition of perovskite batteries. The device can not only move horizontally, but also the fixed frame is connected to the drive device, and can also move longitudinally. By moving longitudinally, the change in resistance value is recorded to determine the location of the area on the component where the break point is not broken. While the aforementioned probe resistance method fills the gap in perovskite cell P1 line detection, it still has the following drawbacks: 1. The probe needs to be in direct contact with the surface of the perovskite cell. The transparent electrode layer (such as ITO) is brittle and thin, and mechanical contact may cause surface scratches or electrode layer peeling, reducing cell efficiency. 2. It requires driving a moving mechanism to detect resistance line by line, which is inefficient and cannot be matched with high-speed laser etching production lines. 3. The resistance method can only determine whether the line is electrically connected, but cannot detect defects such as debris, edge burrs, and line width deviations.

[0005] Patent document CN112599638A discloses a laser line positioning system and method. Light emitted by a light source passes through a cell and enters a photosensitive recognition device to obtain the position of the P1 etching line on the photosensitive recognition device. The relative positional relationship between the position of the P1 etching line on the photosensitive recognition device and the zero point is then calculated to achieve positioning of the laser line. This invention can replace manual experience. When laser etching the P2 and P3 lines, the line position is identified based on the P1 line. The invention can effectively reduce line spacing and dead zones, and can also judge the etching effect and adjust the etching parameters in a timely manner to avoid over- and under-etching problems, thereby improving the yield rate. However, the above-mentioned transmitted light detection method achieves line positioning and over- / under-etching judgment through transmitted light intensity and spectral analysis. Although it is superior to traditional manual experience methods in positioning accuracy and process adjustment, and has achieved automation breakthroughs in line positioning and basic process control, its defects such as transmission dependence, insufficient sensitivity, and speed bottlenecks limit its application in the mass production of high-end perovskites. Summary of the Invention

[0006] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a method and system for online detection of laser perovskite cell etching quality, which realizes non-contact, real-time detection of submicron defects through high-resolution optical imaging and image processing technology, and cooperates with laser etching equipment to form a "processing-detection-correction" closed-loop control to improve etching accuracy and efficiency.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect of the present invention, a method for online detection of the quality of laser perovskite cell etching is provided, wherein a first CCD camera assembly and a second CCD camera assembly are respectively mounted on the left and right sides of a laser processing head, and the first CCD camera assembly and the second CCD camera assembly are respectively aligned with the first and last light beams emitted by the laser processing head;

[0009] Turn on the first and second CCD camera components, perform internal calibration on the CCD camera components, establish the mapping relationship between the pixel coordinate system and the world coordinate system, and calculate the pixel equivalent, which is recorded as Scalar ;

[0010] Drive the laser processing head to move along the positive direction of the X axis, and turn on the laser to perform line processing on the workpiece;

[0011] While the laser processing head is moving forward, the first CCD camera component is started synchronously to collect images, and the images are transmitted to the host computer for real-time image processing to obtain quality analysis results, wherein image collection and image processing are parallel processing;

[0012] The host computer adjusts the laser power in real time, compensates for the marking position, and records the coordinates of the workpiece defect position based on the quality analysis results;

[0013] When the laser processing head reaches the end of the X-axis travel, the laser is turned off, and the laser processing head is driven to move in a stepping motion along the positive direction of the Y-axis for a specified distance, then moves in the negative direction of the X-axis and restarts the laser, and the synchronous trigger signal source is switched to the second CCD camera assembly;

[0014] While the laser processing head moves in the reverse direction, the second CCD camera assembly is synchronously triggered to capture images and repeat the above image processing and real-time adjustment processes;

[0015] During the line engraving process, as the laser processing head moves back and forth in the forward and reverse directions, the first CCD camera assembly and the second CCD camera assembly operate alternately. The dual-camera alternating working mode eliminates the field of view obstruction caused by the movement direction.

[0016] Furthermore, since the laser energy is Gaussian distributed, when the perovskite cell is scribed, defects such as the overall offset of the upper and lower edges, excessively thin or wide line widths due to inaccurate laser power, and burrs on the scribed line edges may occur. Therefore, the specific process of image processing the images captured by the first CCD camera assembly and the second CCD camera assembly is as follows:

[0017] The original image is cropped with a ROI; the Otsu method is used to adaptively calculate the binarization threshold of the cropped ROI image to generate a binary image; the morphological opening operation is used to remove irrelevant information in the binary image; the edge-finding logic algorithm is then used to find the two black and white intersection edges of the engraved line, and two edge lines and one center line are fitted, a total of three lines; the engraved line quality analysis is performed using these three lines to obtain the engraved line quality analysis results.

[0018] Furthermore, the local threshold segmentation method is used to divide the captured ROI image into several sub-regions, and the Otsu method is applied independently in each sub-region to calculate the local threshold, and finally the global binarization result is generated by interpolation.

[0019] Furthermore, the edge-finding logic algorithm is used to find the two black and white intersection edges of the engraved line. The specific process of fitting two edge lines and a center line is as follows:

[0020] (1) Traverse the first column and find the first white area after the first black area from top to bottom. The first pixel of this area is recorded as the starting point of the white area. ;

[0021] (2) Continue to search for the first black area after this white area, and record the last pixel of this white area as the end point of the white area ;

[0022] (3) Calculate the midpoint of the white area in this column, that is:

[0023] ;

[0024] (4) Repeat the above steps to obtain the upper edge point set of the engraved line The set of points on the lower edge of the score line , and get the set of midpoints of the scale lines ;

[0025] (5) Use the least squares method to calculate the upper and lower edge points of the engraved line. , the lower edge point set of the scale line 、Collection of midpoints of engraved lines Perform straight line fitting respectively to obtain two edge lines and a center line.

[0026] Furthermore, the three lines are used to perform line quality analysis, and the specific process of obtaining the quality analysis results includes: calculating the difference between the actual width value and the set width value between the two edge lines in the current real-time image, and judging whether the real-time power of the laser is matched based on the size of the difference between the actual width value and the set width value.

[0027] Specifically, the calculation formula for the actual width value between two edge lines is as follows:

[0028] ;

[0029] in and are the y coordinates of the upper and lower edges of the i-th column scale line, respectively. Scalar is the pixel equivalent, N is the total number of columns in the current real-time image, and Width is the actual width between two edge lines in the current real-time image.

[0030] Furthermore, the three lines are used to perform line quality analysis, and the specific process of obtaining the quality analysis results also includes: calculating the Y-axis offset between the centerline coordinates in the current real-time image and the reference centerline coordinates, and judging whether the Y-axis position of the workpiece needs to be adjusted based on the Y-axis offset.

[0031] Specifically, the specific process of calculating the Y-axis offset between the centerline coordinates in the current real-time image and the reference centerline coordinates is as follows:

[0032] Manually input or shoot the image coordinate system line equation for calculating the reference theoretical path ;

[0033] Real-time detection of the center line coordinates of the scale ;

[0034] The current coordinates Substitute the line equation , get the theoretical image coordinates in the theoretical Y direction ;

[0035] Calculate the Y-axis offset , the formula is as follows:

[0036] ;

[0037] in is the Y coordinate of the midpoint of the currently detected scale line, are theoretical image coordinates, Scalar is the pixel equivalent;

[0038] When the Y axis offset If it is greater than the set value, it means that the Y-axis position error of the workpiece needs to be compensated.

[0039] Furthermore, the three lines are used to perform line quality analysis, and the specific process of obtaining the quality analysis results also includes:

[0040] The slope difference of the straight lines fitted by the two edge lines is calculated. If the slope difference is greater than the set threshold, it means that the marking line is unqualified. The actual workpiece defect position coordinates are calculated and recorded based on the image timestamp and movement speed.

[0041] A second aspect of the present invention provides an online detection system for laser perovskite cell etching quality for implementing the above-mentioned online detection method, comprising a laser processing head, a first CCD camera assembly, a second CCD camera assembly, a gantry, a workpiece, a machine tool, and a displacement platform;

[0042] The machine tool is equipped with a displacement platform and a gantry. A workpiece is placed on the displacement platform, and the displacement platform can drive the workpiece to move along the Y-axis. The gantry is equipped with a laser processing head. The laser processing head has an optical path formed by multiple optical devices, emitting N light beams. The gantry can drive the laser processing head to move along the X-axis and Z-axis. A first CCD camera assembly and a second CCD camera assembly are respectively mounted on either side of the laser processing head along the X-axis. The first and second CCD camera assemblies are aligned with the first and last light beams emitted by the laser processing head, respectively, so that they can constantly capture the processed scribed lines as they move along the X-axis. The CCD camera assembly is equipped with a continuously variable magnification eyepiece and a focusing objective lens, so that the camera imaging can clearly capture the width of a single laser-etched line.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] (1) The present invention adopts a dual-camera alternating working mode, and installs high-resolution CCD camera components on both sides of the laser processing head in the X direction to eliminate occlusion in the direction of movement. In conjunction with the motion mechanism, high-speed real-time online detection is achieved to match the rhythm of the laser etching production line.

[0045] (2) The submicron optical imaging of the CCD camera component is combined with the image processing algorithm to analyze the line quality. A local adaptive threshold segmentation is proposed. The Otsu method is used to adaptively calculate the binarization threshold. The morphological opening operation is then used to remove irrelevant information. The edge-finding logic algorithm is then used to find the two edges of the line. Finally, two edges + a center line are fitted. The line quality analysis is performed through these three lines. The line quality is detected to perform real-time power adjustment (dynamically adjust the laser power according to the line width deviation), position calibration (adjust the Y-axis platform position according to the center line offset, set the dead zone to avoid small jitter interference), and mark the defect position (record the coordinates of the defect area with a large slope difference for subsequent repair). The high-resolution optical detection system has a fast calculation speed and can perform detection while processing, realizing the "processing-detection-correction" closed-loop control. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] Figure 1 This is a schematic diagram of the structure of the online detection system for laser perovskite cell etching quality in the present invention;

[0048] Figure 2 This is a schematic structural diagram of the laser processing head in the present invention;

[0049] Figure 3 Schematic diagram of the position of the laser marking line and CCD camera components in the present invention;

[0050] Figure 4 The original image captured by the CCD camera assembly of the present invention;

[0051] Figure 5 The ROI image is obtained by performing ROI interception on the original image in the present invention;

[0052] Figure 6 The binarized image is obtained by binarizing the intercepted ROI image in the present invention;

[0053] Figure 7 is the image after morphological opening operation in the present invention;

[0054] Figure 8 This is a diagram showing the calculation results of the midpoint of the white scribed line in the present invention;

[0055] Figure 9 This is the least squares straight line fitting result diagram in the present invention;

[0056] Among them, the specific drawings are marked as follows:

[0057] Laser processing head 1, first CCD camera assembly 2, second CCD camera assembly 3, gantry 4, workpiece 5, machine tool 6, displacement platform 7. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] Example 1

[0060] This embodiment discloses a method for online detection of laser perovskite cell etching quality. Figures 1 to 3 As shown, a first CCD camera assembly 2 and a second CCD camera assembly 3 are respectively installed on the left and right sides of the laser processing head 1, and the first CCD camera assembly 2 and the second CCD camera assembly 3 are respectively aligned with the first light beam and the last light beam emitted by the laser processing head 1;

[0061] The entire online detection method process is as follows:

[0062] (1) Turn on the first CCD camera component 2 and the second CCD camera component 3, use Zhang Zhengyou calibration method to calibrate the internal parameters of the CCD camera components, establish the mapping relationship between the pixel coordinate system and the world coordinate system, and calculate the pixel equivalent, which is recorded as Scalar (μm / pixel);

[0063] (2) Drive the laser processing head 1 to move at a constant speed along the positive direction of the X axis, and at the same time turn on the laser (pulsed laser) to perform P1 line processing on the workpiece 5;

[0064] (3) While the laser processing head 1 is moving forward, the motion platform encoder generates a synchronous trigger signal to start the first CCD camera component 2 to capture images, which are then transmitted to the host computer for real-time image processing to obtain quality analysis results. The image acquisition and image processing are processed in parallel, achieving the effect of "shooting and calculating at the same time";

[0065] (4) The host computer adjusts the laser power in real time, compensates for the marking position (motion platform position compensation), and records the coordinates of the workpiece defect position based on the quality analysis results;

[0066] (5) When the laser processing head 1 reaches the end of the X-axis travel, the laser is turned off, and the laser processing head 1 is driven to move in a stepping manner along the positive direction of the Y-axis by a specified distance (generally the line spacing × N), and then moves in the negative direction of the X-axis and restarts the laser, and the synchronous trigger signal source is switched to the second CCD camera assembly 3;

[0067] (6) When the laser processing head 1 moves in the reverse direction, the reverse motion encoder synchronously triggers the second CCD camera assembly 3 to capture images and repeat the above image processing and real-time adjustment processes;

[0068] (7) During the line engraving process, as the laser processing head 1 moves back and forth in the forward and reverse directions, the first CCD camera assembly 2 and the second CCD camera assembly 3 operate alternately. The dual-camera alternating working mode eliminates the field of view obstruction caused by the movement direction.

[0069] The specific process of performing image processing on the images captured by the first CCD camera assembly 2 and the second CCD camera assembly 3 in step (3) and step (6) is as follows:

[0070] 1. The image collected by the CCD camera component is as follows: Figure 4 As shown in the figure, since the coaxial light source illuminates the surface of the target perovskite cell with a Gaussian distribution, it is necessary to perform ROI interception on the original image to obtain an area with uniform illumination and good imaging quality, as shown in the figure below. Figure 5 ROI images are shown.

[0071] 2. Next, the cropped ROI image needs to be binarized. Although the cropped ROI image improves image quality, the illumination is still uneven. The traditional global Otsu method may fail to adapt to local contrast variations, causing segmentation failure. To this end, this embodiment uses local threshold segmentation: the image is divided into several subregions, and the Otsu method is independently applied to each subregion to calculate the local threshold. Finally, the global binarization result is generated through interpolation.

[0072] The specific steps are as follows:

[0073] (1) Divide the image into M×N sub-blocks, each block is w × w ;

[0074] (2) For each sub-block , calculate the Otsu threshold ;

[0075] (3) In order to avoid threshold jumps between sub-blocks, bilinear interpolation is used to generate a smooth threshold surface ;

[0076] (4) Compare the original image pixel by pixel and , generate a binary image, that is:

[0077] ;

[0078] in is the output pixel, and the segmented binary image is as follows Figure 6 shown.

[0079] 3. Use the morphological opening operation, which is mathematically defined as:

[0080] ;

[0081] Where A is the input image, B is the structural element, C is the output image, ⊖ represents the erosion operation, and ⊕ represents the dilation operation. The opening operation can effectively eliminate small noise points and smooth the edges. Figure 7 shown.

[0082] The structuring element selected for the morphological opening operation here is a rectangle. Rectangles are usually used to process horizontal or vertical noise, or to preserve the features of right-angled edges. Since there is a lot of large particle noise in the image, the size of the structuring element is selected to be 15×15. The size selection needs to be adjusted according to the actual scenario. Figure 7 It can be concluded that the morphological opening operation can retain the characteristics of the main lines.

[0083] 4. Use the edge-finding logic algorithm to find the edges where the two black and white areas meet, and fit the center line of the white area (i.e., the engraved line). The specific logic is as follows:

[0084] (1) Traverse the first column and find the first white area after the first black area from top to bottom. The first pixel of this area is recorded as the starting point of the white area. ;

[0085] (2) Continue to search for the first black area after this white area, and record the last pixel of this white area as the end point of the white area ;

[0086] (3) Calculate the midpoint of the white area in this column, that is

[0087] ;

[0088] (4) Repeat the above steps to obtain the upper edge point set and the lower edge point set , get the set of midpoints of the scale lines ,like Figure 8 , where the blue points are the edge points of the scale line and , the red dots are the midpoints of the scale lines ;

[0089] (5) Use the least squares method to calculate the upper and lower edge points of the engraved line. , the lower edge point set of the scale line 、Collection of midpoints of engraved lines Perform straight line fitting respectively to obtain two edge lines and a center line, such as Figure 9 shown.

[0090] Since the laser energy is Gaussian distributed, when marking the perovskite cell, defects such as the overall offset of the upper and lower edges, inaccurate laser power resulting in line width that is too thin or too wide, and burrs on the edge of the marking line may occur. Therefore, in step (3), the marking quality analysis is performed using these three lines, which mainly includes three aspects (the actual width value between the two edge lines, the Y-axis offset of the centerline coordinate, and the slope difference between the two edge lines). In step (4), the host computer adjusts the laser power in real time, compensates the marking position (motion platform position compensation), and records the coordinates of the workpiece defect position according to the quality analysis results.

[0091] (1) Calculate the difference between the actual width and the set width between the two edge lines in the current real-time image, and judge whether the real-time power of the laser is matched based on the difference between the actual width and the set width. If the actual width is less than the set width, increase the laser power; otherwise, decrease the laser power.

[0092] Specifically, the actual width between the two edge lines is calculated by multiplying the Y coordinate difference between the upper and lower edge points of the current captured image by the pixel equivalent to obtain the displayed width. The calculation formula is as follows:

[0093] ;

[0094] in and are the y coordinates of the upper and lower edges of the i-th column scale line, respectively. Scalar is the pixel equivalent, N is the total number of columns in the current real-time image, and Width is the actual width between two edge lines in the current real-time image.

[0095] (2) Calculate the Y-axis offset between the centerline coordinates in the current real-time image and the reference centerline coordinates, and determine whether the Y-axis position of workpiece 5 needs to be adjusted based on the Y-axis offset. Note that a dead zone value needs to be set to ignore small jitters (<±1μm).

[0096] The specific steps are as follows:

[0097] 1. Manually input or photograph the image coordinate system line equation for the calculation benchmark theoretical path ;

[0098] 2. Real-time detection of the center line coordinates of the engraved lines ;

[0099] 3. Set the current coordinates Substitute the line equation , get the theoretical image coordinates in the theoretical Y direction ;

[0100] 4. Calculate the Y-axis offset , the formula is as follows:

[0101] ;

[0102] in is the Y coordinate of the midpoint of the currently detected scale line, are theoretical image coordinates, Scalar is pixel equivalent.

[0103] When the Y axis offset If it is greater than the set value, for example >±1 μm, it means that the Y-axis position of the workpiece 5 needs to be adjusted to compensate for the error.

[0104] (3) Calculate the slope difference of the straight lines fitted by the two edge lines. If the slope difference is greater than the set threshold, it means that the marking line is unqualified. The actual workpiece defect position coordinates are calculated and recorded based on the image timestamp and movement speed.

[0105] Example 2

[0106] This embodiment discloses a laser perovskite cell etching quality online detection system for realizing the above-mentioned online detection method. Figures 1 to 3 , including a laser processing head 1, a first CCD camera assembly 2, a second CCD camera assembly 3, a gantry 4, a workpiece 5, a machine tool 6 and a displacement platform 7;

[0107] The machine tool 6 is mounted with a displacement platform 7 and a gantry 4. A workpiece 5 is placed on the displacement platform 7, which can move the workpiece 5 along the Y-axis. The gantry 4 is mounted with a laser processing head 1. Multiple optical components form an optical path within the laser processing head 1, emitting N light beams. The gantry 4 can move the laser processing head 1 along the X- and Z-axes. A first CCD camera assembly 2 and a second CCD camera assembly 3 are mounted on either side of the laser processing head 1 along the X-axis. The first and second CCD camera assemblies 2 and 3 are aligned with the first and last light beams emitted by the laser processing head 1, respectively, so that they can capture the processed scribed lines at all times as they move along the X-axis. The CCD camera assembly is equipped with a continuously variable magnification eyepiece and a focusing objective lens, enabling the camera to clearly capture the width of a single laser-etched line. In this solution, the CCD camera assembly uses a 20x magnification focusing objective lens combined with a 2x eyepiece, paired with a 280w industrial camera with a maximum frame rate of 132fps.

[0108] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for online detection of laser perovskite cell etching quality, characterized in that: A first CCD camera assembly and a second CCD camera assembly are respectively installed on the left and right sides of the laser processing head, and the first CCD camera assembly and the second CCD camera assembly are respectively aligned with the first light beam and the last light beam emitted by the laser processing head; Turn on the first and second CCD camera components, perform internal calibration on the CCD camera components, establish the mapping relationship between the pixel coordinate system and the world coordinate system, and calculate the pixel equivalent, which is recorded as Scalar ; Drive the laser processing head to move along the positive direction of the X axis, and turn on the laser to perform line processing on the workpiece; While the laser processing head is moving forward, the first CCD camera component is started synchronously to collect images, and the images are transmitted to the host computer for real-time image processing to obtain quality analysis results, wherein image collection and image processing are parallel processing; The host computer adjusts the laser power in real time, compensates for the marking position, and records the coordinates of the workpiece defect position based on the quality analysis results; When the laser processing head reaches the end of the X-axis travel, the laser is turned off, and the laser processing head is driven to move in a stepping motion along the positive direction of the Y-axis for a specified distance, then moves in the negative direction of the X-axis and restarts the laser, and the synchronous trigger signal source is switched to the second CCD camera assembly; While the laser processing head moves in the reverse direction, the second CCD camera assembly is synchronously triggered to capture images and repeat the above image processing and real-time adjustment processes; During the line engraving process, as the laser processing head moves back and forth in the forward and reverse directions, the first CCD camera assembly and the second CCD camera assembly operate alternately.

2. The online detection method for laser perovskite cell etching quality according to claim 1, characterized in that: The specific process of performing image processing on the images collected by the first CCD camera assembly and the second CCD camera assembly is as follows: The original image is cropped with a ROI; the Otsu method is used to adaptively calculate the binarization threshold of the cropped ROI image to generate a binary image; the morphological opening operation is used to remove irrelevant information in the binary image; the edge-finding logic algorithm is then used to find the two black and white intersection edges of the engraved line, and two edge lines and one center line are fitted, a total of three lines; the engraved line quality analysis is performed using these three lines to obtain the engraved line quality analysis results.

3. The online detection method for laser perovskite cell etching quality according to claim 2, characterized in that: The local threshold segmentation method is used to divide the intercepted ROI image into several sub-regions. The Otsu method is applied independently in each sub-region to calculate the local threshold, and finally the global binarization result is generated by interpolation.

4. The online detection method for laser perovskite cell etching quality according to claim 3, characterized in that: The specific process of using the edge-finding logic algorithm to find the two black and white intersection edges of the engraved line and fitting two edge lines and a center line is as follows: (1) Traverse the first column and find the first white area after the first black area from top to bottom. The first pixel of this area is recorded as the starting point of the white area. ; (2) Continue to search for the first black area after this white area, and record the last pixel of this white area as the end point of the white area ; (3) Calculate the midpoint of the white area in this column, that is: ； (4) Repeat the above steps to obtain the upper edge point set of the engraved line The set of points on the lower edge of the score line , and get the set of midpoints of the scale lines ; (5) Use the least squares method to calculate the upper and lower edge points of the engraved line. , the lower edge point set of the scale line 、Collection of midpoints of engraved lines Perform straight line fitting respectively to obtain two edge lines and a center line.

5. The online detection method for laser perovskite cell etching quality according to claim 2, characterized in that: The three lines are used to analyze the line quality. The specific process of obtaining the quality analysis results includes: calculating the difference between the actual width value and the set width value between the two edge lines in the current real-time image, and judging whether the real-time power of the laser is matched based on the difference between the actual width value and the set width value.

6. The online detection method for laser perovskite cell etching quality according to claim 5, characterized in that: The actual width between two edge lines is calculated as follows: ； in and are the y coordinates of the upper and lower edges of the i-th column scale line, respectively. Scalar is the pixel equivalent, N is the total number of columns in the current real-time image, and Width is the actual width between two edge lines in the current real-time image.

7. The online detection method for laser perovskite cell etching quality according to claim 5 or 6, characterized in that: The quality analysis of the engraving is performed using these three lines. The specific process of obtaining the quality analysis results also includes: calculating the Y-axis offset between the centerline coordinates in the current real-time image and the reference centerline coordinates, and judging whether the Y-axis position of the workpiece needs to be adjusted based on the Y-axis offset.

8. The online detection method for laser perovskite cell etching quality according to claim 7, characterized in that: The specific process of calculating the Y-axis offset between the centerline coordinates in the current real-time image and the reference centerline coordinates is as follows: Manually input or shoot the image coordinate system line equation for calculating the reference theoretical path ; Real-time detection of the center line coordinates of the scale ; The current coordinates Substitute the line equation , get the theoretical image coordinates in the theoretical Y direction ; Calculate the Y-axis offset , the formula is as follows: ； in is the Y coordinate of the midpoint of the currently detected scale line, are theoretical image coordinates, Scalar is the pixel equivalent; When the Y axis offset If it is greater than the set value, it means that the Y-axis position error of the workpiece needs to be compensated.

9. The online detection method for laser perovskite cell etching quality according to claim 8, characterized in that: The specific process of analyzing the quality of the line through these three lines and obtaining the quality analysis results also includes: The slope difference of the straight lines fitted by the two edge lines is calculated. If the slope difference is greater than the set threshold, it means that the marking line is unqualified. The actual workpiece defect position coordinates are calculated and recorded based on the image timestamp and movement speed.

10. An online detection system for laser perovskite cell etching quality for implementing the online detection method according to any one of claims 1 to 9, characterized in that: It includes a laser processing head, a first CCD camera assembly, a second CCD camera assembly, a gantry, a workpiece, a machine tool and a displacement platform; A displacement platform and a gantry are installed on the machine tool. A workpiece is placed on the displacement platform, and the displacement platform can drive the workpiece to move along the Y-axis direction. A laser processing head is installed on the gantry, and the gantry can drive the laser processing head to move along the X-axis and Z-axis directions. A first CCD camera assembly and a second CCD camera assembly are respectively installed on the left and right sides of the laser processing head. The first CCD camera assembly and the second CCD camera assembly are respectively aligned with the first light beam and the last light beam emitted by the laser processing head.

Images