A novel white light interferometry method for detecting thin film morphology in display inkjet printing manufacturing

Abstract

The present application belongs to the field of novel display jet printing manufacturing, and particularly relates to a white light interference detection method for film morphology in novel display jet printing manufacturing, which comprises: obtaining a series of white light scanning interference images of a to-be-detected region; adaptively determining whether the to-be-detected region is a film region or an edge region by using an apex angle method; for the film region, constructing discrete points by using a series of light intensity contrasts of each pixel point and a series of image acquisition heights, performing envelope extraction and Gaussian fitting, determining the image acquisition height of a double-peak in the fitting signal of the corresponding pixel point, and calculating the absolute film thickness at the corresponding pixel point; for the edge region, adaptively dividing a film area and a substrate area based on envelope extraction and Gaussian fitting, using the image acquisition height of the higher peak in the fitting signal of each pixel point in the film area and the image acquisition height of the single peak in the fitting signal of a certain pixel point on the substrate, and calculating the relative film thickness of the corresponding pixel point in the film area. The method can adaptively realize accurate morphology reconstruction of different film regions.

Description

Technical Field

[0001] This invention belongs to the field of novel display inkjet printing manufacturing, and more specifically, relates to a white light interferometry detection method for thin film morphology in novel display inkjet printing manufacturing. Background Technology

[0002] With the widespread application of inkjet printing technology in the manufacturing of new display devices, the demand for quality inspection and control of inkjet-printed films during the manufacturing process is increasing.

[0003] In the inkjet printing manufacturing process of new display devices, thin-film encapsulation and inkjet printing are indispensable steps. The quality of the inkjet-printed thin film directly affects the encapsulation effect of the display device, and is directly related to the display device's performance characteristics such as light emission uniformity and resistance to water and oxygen corrosion.

[0004] For the inspection of printed thin films in the manufacturing process of new display devices, the inspection accuracy needs to reach the micrometer, submicrometer, or even nanometer level. Some traditional inspection methods on the production line include various point thickness sensors and step meters. Among them, many point thickness sensors can only detect the film thickness data at one point at a time, while step meters are a contact inspection method, which is not conducive to controlling inspection costs and improving inspection efficiency.

[0005] White light interferometry, due to its advantages of high precision, non-contact operation, and surface imaging, has been attempted for use in the inspection of inkjet-printed thin films in the manufacturing process of new display devices. However, when the substrate has high reflectivity, the scanning interference signal in the thin film area will exhibit a bimodal characteristic, while the scanning interference signal in the substrate area will exhibit a unimodal characteristic. Therefore, there will be interference signals with multiple characteristics within the current field of view. However, general white light scanning interferometry algorithms only set the white light scanning interference signal within the current field of view to a unimodal or bimodal measurement mode. Therefore, this simple and crude method will produce errors in the reconstruction of the thin film edge morphology, meaning that the accuracy of film thickness detection cannot be guaranteed.

[0006] Therefore, in order to meet the white light interferometry detection requirements of thin film morphology in novel display inkjet printing manufacturing, it is urgent to propose a white light interferometry detection method that can detect both the thin film area and the thin film edge area, and that can adapt to various interference signal characteristics within the field of view when detecting the thin film edge area. Summary of the Invention

[0007] To address the shortcomings and improvement needs of existing technologies, this invention provides a novel white light interferometry detection method for thin film morphology in inkjet printing manufacturing. The purpose is to propose a white light interferometry detection method that can detect both the thin film region and the thin film edge region, and that adapts to multiple features within the field of view when detecting the thin film edge region, thereby enabling accurate morphology reconstruction of different thin film regions adaptively.

[0008] To achieve the above objectives, according to one aspect of the present invention, a novel white light interferometry method for detecting thin film morphology in inkjet printing manufacturing is provided, comprising:

[0009] Acquire a series of white light scanning interferometric images of the area to be inspected on the printed thin film;

[0010] Extract the series light intensity contrast of any three vertex pixels in the series of images, and perform envelope extraction and Gaussian fitting on the series light intensity contrast of each vertex. If the fitted signals of the three vertex pixels are all bimodal signals, the area to be detected is determined to be a thin film region; otherwise, it is determined to be a thin film edge region.

[0011] When the region is a thin film area, discrete points are constructed using the series of light intensity contrasts and corresponding image acquisition heights of each pixel within the region. This is used for envelope extraction and Gaussian fitting, and the image acquisition height at the bimodal peak in the fitted signal of the corresponding pixel is determined to calculate the film thickness at that pixel. When the region is a thin film edge area, the pixels in this area are divided into film and substrate areas based on envelope extraction and Gaussian fitting. Discrete points are constructed using the series of light intensity contrasts and corresponding image acquisition heights of each pixel within the film area. This is used for envelope extraction and Gaussian fitting, and the image acquisition height h at the higher peak in the fitted signal of the corresponding pixel is determined. Bu Discrete points are constructed using a series of light intensity contrasts and corresponding image acquisition heights of a pixel within the substrate area to perform envelope extraction and Gaussian fitting, thereby determining the image acquisition height h at the single peak in the fitted signal of that pixel. A Calculate each h Bu and h A The difference is used as the film thickness of the corresponding pixel in the film region to complete the detection.

[0012] Furthermore, the series light intensity contrast of each pixel is obtained by normalizing and denoising the series gray values ​​of that pixel.

[0013] Furthermore, when it is a thin film region, discrete points are constructed by using the series of light intensity contrasts of each pixel in the region and the corresponding series of image acquisition relative heights relative to the acquisition position of the first frame interference image, in order to perform the envelope extraction and Gaussian fitting. The relative height of each series of image acquisitions is calculated by multiplying the frame number of the corresponding scan interference image with the white light interference scan step.

[0014] Then the film thickness h at each pixel point within the region r =(h u -h d ) / n, where n is the refractive index of the thin film, h u h d The relative heights at which images were acquired at the bimodal locations are shown.

[0015] Furthermore, using Hilbert transform, the envelope extraction is performed on discrete points constructed based on a series of light intensity contrasts of the target pixels;

[0016] The Gaussian fitting specifically involves performing nonlinear iterative fitting on the extracted envelope based on a Gaussian function model to obtain the Gaussian model of the fitted signal for that pixel.

[0017] Furthermore, the Gaussian function model is expressed as:

[0018]

[0019] In the formula, B(t) is the fitted signal; N represents the number of peaks, which takes the value 0 or 1; H i g(p) represents the amplitude corresponding to the i-th Gaussian component. i ,w i Let p be the center position of the i-th Gaussian component. i Width is w i The Gaussian function expression;

[0020] The limiting condition for the nonlinear iterative fitting is:

[0021]

[0022] In the formula, E is the weighted sum of squared errors of the iterative fitting, and w i The weights of the i-th discrete points in the envelope when fitting the envelope; Let y be the light intensity contrast value of the i-th envelope discrete point; i h is the fitted value of the light intensity contrast at the i-th envelope discrete point; g h is the weighted center height calculated using the centroid method based on the discrete points obtained from the above construction. i The image acquisition height is denoted by l; ζ is the adjustment coefficient; l c I is the coherence length of the white light source; ip Let be the light intensity contrast of the i-th discrete point obtained by the construction.

[0023] Furthermore, when performing envelope extraction and Gaussian fitting on the series of light intensity contrasts at each vertex, or when dividing the pixel points of the area to be detected into film region and substrate region based on envelope extraction and Gaussian fitting, the nonlinear iterative fitting specifically refers to:

[0024] The number of Gaussian functions in the Gaussian function model is set to 1, and the extracted envelope is subjected to nonlinear iterative fitting to obtain the fitting discrete points of the Gaussian function. Based on the fitting discrete points and the discrete points in the extracted envelope, the root mean square error RMS0 and the coefficient of determination R-squared0 of the iterative fitting are calculated.

[0025] The number of Gaussian functions in the Gaussian function model is set to 2, and the extracted envelope is subjected to nonlinear iterative fitting to obtain the fitting discrete points after the superposition of the two Gaussian functions. Based on the fitting discrete points and the discrete points in the extracted envelope, the root mean square error RMS1 and the coefficient of determination R-squared1 of the iterative fitting are calculated.

[0026] When both R-squared0 and R-squared1 are greater than preset values, the number N of Gaussian functions is determined according to the formula; otherwise, nonlinear iterative fitting is performed again to finally obtain the fitted signal Gaussian model of the target pixel. The formula is as follows: In the formula, when N=0, it means that the fitted signal corresponding to the target pixel is a single-peak signal; when N=1, it means that the fitted signal corresponding to the target pixel is a double-peak signal.

[0027] Furthermore, the method for dividing the film region and the substrate region is as follows:

[0028] (1) Based on whether the fitted signal corresponding to any three vertex corners is a bimodal signal or a unimodal signal, a label value m is used for labeling. The m value is 1 for bimodal signals and 0 for unimodal signals.

[0029] (2) If the m values ​​of the two vertex corners in the same column are different, the coordinates of the first column pixel point are extracted; if the m values ​​of the two vertex corners in the same row are different, the coordinates of the first row pixel point are extracted; and the range of the dimension coordinate v that changes in the extracted series of pixel point coordinates is determined to be 0 to R-1 or 0 to C-1, where R is the total number of rows in the interference image and C is the total number of columns in the interference image;

[0030] (3) Assign 0 to a, and assign R-1 or C-1 to b, and extract respectively. The series of light intensity contrasts of each pixel are obtained, and after envelope extraction and Gaussian function fitting, the corresponding marker values ​​of the three pixels are obtained. In this case, int() represents integer division;

[0031] (4) If m aand Different values, then With m b If the values ​​are the same, The marker value of each pixel within the range The values ​​are all set to 0. and the current Assign the value of m to b, and repeat step (3); if m a and If they are the same, then With m b Different values ​​will The marker value corresponding to each pixel within the range The value of each is set to the current value of m. a The value to be determined, and the current value to be determined. The value of is assigned to 'a', and step (3) is repeated.

[0032] (5) Steps (3) and (4) are executed independently for each other row or column in the area to be detected to obtain the m value corresponding to each pixel in the area to be detected. Based on the m value, the film area and the substrate area are divided.

[0033] Furthermore, it also includes: evaluating the film forming quality using the calculated film thickness at each pixel in the region to be detected.

[0034] The present invention also provides a novel white light interferometry detection device for displaying the morphology of thin films in inkjet printing manufacturing, comprising: a white light scanning interferometry module and a processor;

[0035] The white light scanning interferometry module is used to acquire a series of white light scanning interferometry images of the area to be inspected in the inkjet printing film; the processor executes the novel white light interferometry detection method for displaying the morphology of the thin film in inkjet printing manufacturing as described above.

[0036] The present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed by a processor, it controls the device containing the storage medium to perform a novel white light interferometry detection method for thin film morphology in display inkjet printing manufacturing as described above.

[0037] In summary, the above-described technical solutions conceived in this invention can achieve the following beneficial effects:

[0038] (1) This invention proposes an envelope extraction and Gaussian fitting method to identify whether a pixel in the detection area is located in the thin film region or the thin film edge region. In the specific execution of the method, firstly, based on envelope extraction and Gaussian fitting, combined with the vertex method, the current detection area is automatically identified as either a thin film region or a thin film edge region, so as to apply different film thickness calculation modes. Furthermore, if the current detection area is automatically identified as a thin film edge region by the vertex method, then when calculating the film thickness of the thin film edge region, it is proposed to adaptively divide the detection area into a film region and a substrate region based on envelope extraction and Gaussian fitting, thereby ensuring the reconstruction accuracy of the morphology (film thickness) at the thin film edge, which simultaneously contains the composite features of the thin film region and the substrate region. Therefore, this method realizes two adaptive methods in the white light interferometry detection process of inkjet printed thin films, which can avoid the problems existing in the existing white light interferometry detection of inkjet printed thin films, making this method widely applicable to the accurate detection of film thickness in different scenarios and different thin film regions. Furthermore, the invention's process involves performing two adaptive recognitions before film thickness calculation. This avoids the need for two fittings to determine the number of Gaussian functions to identify whether it's a single or double peak when fitting the Gaussian function to the interference signal of each pixel, thus reducing the algorithm's time complexity for reconstructing the morphology of the thin film edge region. Specifically, the "absolute film thickness" calculation mode for the thin film region avoids the substrate warping error that might be introduced when using the "relative film thickness" mode for sampling points far from the uncoated substrate. Additionally, using only the substrate within the current field of view (i.e., the image acquisition height at the peak of the interference signal corresponding to the same pixel on the lower surface of the thin film) as a reference makes measurement more convenient and accurate. The "relative film thickness" mode for the thin film edge region measurement utilizes only the height information of the interference peak on the upper surface of the thin film region, preventing inaccurate identification of the lower surface height information caused by the distortion of the lower surface interference peak due to dispersion. Therefore, the algorithm proposed in this invention can adaptively achieve accurate morphology reconstruction for different thin film regions.

[0039] (2) This invention proposes to use Hilbert transform for envelope extraction. Hilbert transform extraction has a faster computation speed than Fourier transform extraction and is also faster than continuous wavelet transform extraction without generating redundant information. Based on this, a multi-Gaussian function model is further used to fit the extracted envelope, thereby enabling the splitting of stacked multi-interference peaks to quickly and accurately identify the height information of multiple surfaces of the thin film.

[0040] (3) When reconstructing the film thickness in the edge region of the thin film, it is necessary to divide the region into film region and substrate region. In this regard, the present invention proposes to use the film region and substrate dichotomy method to adaptively identify and mark the film region and substrate region in the edge region of the thin film. This method can quickly identify and mark the film region and substrate region in each row or column with low time complexity. Attached Figure Description

[0041] Figure 1 This is a flowchart of a novel white light interferometry detection process for thin film morphology in display inkjet printing manufacturing, provided by an embodiment of the present invention.

[0042] Figure 2 This is a schematic diagram showing the positions of the three vertices in the vertex method provided in this embodiment of the invention;

[0043] Figure 3 This is a schematic diagram of the "relative film thickness" measurement mode provided in an embodiment of the present invention;

[0044] Figure 4 This is a schematic diagram of the "absolute film thickness" measurement mode provided in an embodiment of the present invention;

[0045] Figure 5 This is a schematic diagram of the steps of the HT-Gaussian algorithm provided in an embodiment of the present invention;

[0046] Figure 6 A schematic diagram of the nonlinear iterative fitting process provided in an embodiment of the present invention;

[0047] Figure 7 This is a schematic diagram illustrating the definition of single peaks and double peaks provided in an embodiment of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0049] Example 1

[0050] A novel white light interferometry method for detecting thin film morphology in display inkjet printing manufacturing, such as... Figure 1 As shown, it includes:

[0051] Acquire a series of white light scanning interferometric images of the area to be inspected on the printed thin film;

[0052] Extract the series light intensity contrast of any three vertex pixels in the series of images, and perform envelope extraction and Gaussian fitting on the series light intensity contrast of each vertex. If the fitted signals of the three vertex pixels are all bimodal signals, the area to be detected is determined to be a thin film region; otherwise, it is determined to be a thin film edge region.

[0053] When the region is a thin film area, discrete points are constructed using the series of light intensity contrasts and corresponding image acquisition heights of each pixel within the region. This is used for envelope extraction and Gaussian fitting to determine the image acquisition height at the bimodal peak in the fitted signal of each pixel, thus calculating the film thickness at that pixel. When the region is a thin film edge area, based on envelope extraction and Gaussian fitting, the pixels in this area are divided into film and substrate areas. Discrete points are constructed using the series of light intensity contrasts and corresponding image acquisition heights of each pixel within the film area. This is used for envelope extraction and Gaussian fitting to determine the image acquisition height h at the higher peak in the fitted signal of each pixel. Bu Discrete points are constructed using a series of light intensity contrasts and corresponding image acquisition heights of a pixel within the substrate area to perform envelope extraction and Gaussian fitting, thereby determining the image acquisition height h at the single peak in the fitted signal of that pixel. A Calculate each h Bu and h A The difference is used as the film thickness of the corresponding pixel in the film region to complete the detection.

[0054] This method proposes an envelope extraction and Gaussian fitting approach to identify whether a pixel within the detection region is located in a thin film region or a thin film edge region. In practice, based on envelope extraction and Gaussian fitting, combined with the vertex method, the method automatically identifies whether the current detection region is a thin film region or a thin film edge region, applying different film thickness calculation modes accordingly. Furthermore, if the current detection region is automatically identified as a thin film edge region using the vertex method, then when calculating the film thickness in the thin film edge region, the method proposes an adaptive division of the detection region into a film region and a substrate region based on envelope extraction and Gaussian fitting. This ensures the accuracy of reconstructing the morphology (film thickness) at the thin film edge, which simultaneously contains composite features of both the thin film region and the substrate region. Therefore, this method achieves two adaptive approaches in the white light interferometry detection process for inkjet-printed thin films, avoiding the problems existing in current white light interferometry detection of inkjet-printed thin films. This allows the method to be widely applied to accurate film thickness detection in different scenarios and for different thin film regions. In addition, the process of this invention performs two adaptive recognitions before film thickness calculation, which avoids having to perform two fittings to determine the number of Gaussian functions to identify whether it is a single peak or a double peak when fitting the Gaussian function to the interference signal of each pixel, thus reducing the algorithm time complexity when reconstructing the morphology of the thin film edge region.

[0055] The method employs a vertex-angle adaptive separation of the thin-film region and the thin-film edge region, applying different film thickness calculation methods to these two regions. The thin-film region uses an "absolute film thickness" calculation mode, calculating the film thickness based on the image acquisition height of the interference signal peaks corresponding to the same pixel on both the upper and lower surfaces of the thin film. The thin-film edge region uses a "relative film thickness" calculation mode, calculating the film thickness based on the image acquisition height of the interference signal peaks corresponding to pixels on the upper surface of the thin film and the image acquisition height of the interference signal peaks corresponding to pixels on the substrate. Using the "absolute film thickness" calculation mode for the thin-film region avoids the substrate warping error that might be introduced when using the "relative film thickness" mode for samples far from the uncoated substrate. Furthermore, using only the substrate within the current field of view (i.e., the image acquisition height of the interference signal peak corresponding to the same pixel on the lower surface of the thin film) as a reference makes measurement more convenient and accurate. The "relative film thickness" mode for the thin-film edge region utilizes only the height information of the interference peaks on the upper surface of the thin film region, preventing inaccurate identification of the lower surface height information caused by the distortion of the lower surface interference peaks due to dispersion in the thin-film region. Therefore, the algorithm proposed in this method can adaptively achieve accurate morphology reconstruction of different thin film regions.

[0056] It should be noted that this method is a detection method based on white light scanning interferometry, which can realize non-contact measurement of printed films, and the reconstruction of the morphology of the printed film after each detection is presented as a surface shape.

[0057] Regarding the aforementioned vertex method, such as Figure 2 As shown, if a single interferometric image has R rows and C columns within its field of view, the coordinate position of a pixel in the interferogram is represented by (c, r). Where c = 0, 1, 2, 3…R⁻¹; r = 0, 1, 2, 3…C⁻¹, and the coordinates of any three vertices are, for example, (0, 0), (R⁻¹, 0), and (0, C⁻¹). After envelope extraction and Gaussian fitting of the series of light intensity contrasts at each vertex, bimodal and unimodal patterns can be identified based on the fitted signal. A unimodal pixel is marked with a value m of 0, and a bimodal pixel is marked with a value m of 1, thus obtaining the marked value m. (0,0) m (R-1，0) m (0，C-1) Where the subscript of m represents the pixel coordinates. When the result M of the AND operation of the three marker values ​​is 1, the area of ​​the film within the current measurement field of view is determined to be the interior of the film; if M is 0, the area of ​​the film within the current measurement field of view is determined to be the edge of the film. The above AND operation is expressed as: M = m 0,0 &m R-1，0 &m 0，C-1 .

[0058] Regarding the above-mentioned "relative film thickness" calculation model, such as Figure 3 As shown, in the edge region of the printed film, the space is equivalent to scanning twice. Through envelope extraction and Gaussian fitting, the two height positions h of the upper and lower surfaces of point B in the film region are calculated. Bu h Bd But only keep h Bu To prevent the use of h Bd This addresses the issue of inaccurate identification of the lower surface height information caused by the distortion of the lower surface interference peak due to dispersion in the thin film region. Furthermore, through envelope extraction and Gaussian fitting, the surface height position h of point A in the substrate region is calculated. A Therefore, the actual film thickness h at point B can be calculated. r :h r =h Bu -h A .

[0059] Regarding the above-mentioned "absolute film thickness" calculation model, such as Figure 4 As shown, in the central region of the printed film, spatially it is equivalent to scanning only once. By extracting the envelope and using Gaussian fitting, the interference peak positions of the upper and lower surfaces of the film appear sequentially in the time series are obtained, thereby determining the actual thickness h of the film. r :

[0060] h r =h j / n

[0061] h j =h u -h d

[0062] Among them, h r h represents the actual thickness of the film. j The calculated thickness of the thin film is given by n; the refractive index of the thin film is given by h. u h d These represent the image acquisition heights at the bimodal locations. Figure 4 In the image, A and A′ represent the points on the upper and lower surfaces of the thin film corresponding to a certain pixel, respectively.

[0063] As a preferred embodiment, the aforementioned series of light intensity contrasts for each pixel are obtained by normalizing and denoising the series of gray values ​​of that pixel.

[0064] As a preferred embodiment, when the region is a thin film, discrete points are constructed using a series of light intensity contrasts of each pixel within the region and a series of image acquisition relative heights relative to the acquisition position of the first frame interference image. These discrete points are then used for the aforementioned envelope extraction and Gaussian fitting. The relative height of each series of image acquisitions is calculated by multiplying the frame number of the corresponding scanned interference image by the white light interference scanning step size. The thin film thickness h at each pixel within the region is then...r =(h u -h d ) / n, where n is the refractive index of the thin film, h u h d The relative heights at which images were acquired at the bimodal locations are shown.

[0065] The image acquisition height for one dimension at each discrete point is the relative image acquisition height relative to the acquisition position of the first frame of the interferometric image. The relative image acquisition height is calculated by multiplying the frame number of the corresponding scanned interferometric image with the white light interferometric scanning step size. This method can improve the acquisition efficiency of the image acquisition height while ensuring the accuracy of the calculation.

[0066] The aforementioned envelope extraction and Gaussian fitting can be considered as preferred implementation methods, such as... Figure 5 As shown, the envelope of discrete points constructed based on a series of light intensity contrasts of the target pixel is extracted using Hilbert transform to obtain the envelope of the AC component of the series of light intensity contrasts; the Gaussian fitting is specifically as follows: based on the Gaussian function model, the extracted envelope is subjected to nonlinear iterative fitting to obtain the Gaussian model of the fitted signal of the pixel.

[0067] The white light scanning interference signal can be expressed as: I(t) = I DC (t)+I AC I(t) is the time series of the white light scanning interference signal; I DC (t) represents the DC component of the time series of the white light scanning interference signal; I AC (t) represents the AC component of the time series of the white light scanning interference signal.

[0068] The above-described envelope extraction and Gaussian fitting method can be simply referred to as the HT-Gaussian algorithm. By combining Hilbert envelope extraction and Gaussian function fitting, Hilbert transform extraction offers faster computation speed than Fourier transform extraction, and it is also faster than continuous wavelet transform extraction without generating redundant information. By fitting the extracted envelope using a multi-Gaussian function model, the stacked multi-interference peaks can be separated, accurately identifying the height information of multiple thin film surfaces.

[0069] As a preferred implementation, the above Gaussian function model is expressed as:

[0070]

[0071] In the formula, B(t) is the fitted signal; N represents the number of peaks, which takes the value 0 or 1; H i g(p) represents the amplitude corresponding to the i-th Gaussian component. i ,w iLet p be the center position of the i-th Gaussian component. i Width is w i The Gaussian function expression.

[0072] The weighted least squares method is used to judge the quality of the iterative fitting. The constraints of the above nonlinear iterative fitting are as follows:

[0073]

[0074] In the formula, E is the weighted sum of squared errors of the iterative fitting, and w i The weights of the i-th discrete points in the envelope when fitting the envelope; Let y be the light intensity contrast value of the i-th envelope discrete point; i h is the fitted value of the light intensity contrast at the i-th envelope discrete point; g h is the weighted center height obtained by using the centroid method based on the discrete points constructed above; i The image acquisition height is denoted by l; ζ is the adjustment coefficient; l c I is the coherence length of the white light source; ip The light intensity contrast of the i-th discrete point obtained above is normalized and denoised. It should be noted that the discrete point obtained above is the discrete point constructed during the current envelope extraction and Gaussian fitting.

[0075] Considering the stacking phenomenon of multiple interference peaks (i.e., when the thin film is relatively thin), the signal envelope intensity of the overlapping part is not equal to the sum of the envelope intensities of the two interference peaks due to the destructive effect of the light wave oscillations of the two interference peaks. The aforementioned nonlinear iterative fitting constraint allows for the correct decomposition of the signal envelope even in this case of multiple interference peak stacking, thereby obtaining more accurate interference peak height information.

[0076] This can be used as a preferred implementation method, such as Figure 6 As shown, when performing envelope extraction and Gaussian fitting on the series of light intensity contrasts at each vertex, or when dividing the pixel points in the detection area into film and substrate regions based on envelope extraction and Gaussian fitting, the above nonlinear iterative fitting is specifically as follows:

[0077] The number of Gaussian functions in the Gaussian function model is set to 1, and nonlinear iterative fitting is performed on the extracted envelope to obtain the fitting discrete points of the Gaussian functions. Based on the fitting discrete points and the discrete points in the extracted envelope, the root mean square error RMS0 and the coefficient of determination R-squared0 of the iterative fitting are calculated.

[0078] The number of Gaussian functions in the Gaussian function model is set to 2, and the extracted envelope is subjected to nonlinear iterative fitting to obtain the fitting discrete points after the superposition of the two Gaussian functions. Based on the fitting discrete points and the discrete points in the extracted envelope, the root mean square error RMS1 and the coefficient of determination R-squared1 of the iterative fitting are calculated.

[0079] When both R-squared0 and R-squared1 are greater than the preset values, the number N of Gaussian functions is determined according to the formula; otherwise, nonlinear iterative fitting is performed again to finally obtain the fitted signal Gaussian model of the target pixel. The formula is as follows: In the formula, when N=0, it indicates that the fitted signal corresponding to the target pixel is a single-peak signal; when N=1, it indicates that the fitted signal corresponding to the target pixel is a double-peak signal. For example, the above preset value can be 0.8.

[0080] Envelope extraction and Gaussian fitting can not only accurately identify the height position carried by the single-peak interference signal in the substrate region to reflect the height information of the substrate, but also determine whether the interference signal is bimodal or single-peaked based on the fitting result of the Gaussian function, thereby determining whether the pixel where the interference signal is located is in the thin film region or the substrate region. Figure 7 As shown.

[0081] It should be noted that in the iterative fitting required for film thickness calculation, it is only necessary to set the number of Gaussian functions for pixels in the thin film region to 2 and the number of Gaussian functions for pixels in the edge region to 1. This provides the necessary information for the subsequent adaptive differentiation between the thin film region and the thin film edge region, as well as the adaptive differentiation between the thin film region and the substrate region in the thin film edge region. This enables the algorithm to differentiate between different regions and features, thereby allowing for accurate identification of the height information of different regions of the thin film.

[0082] As a preferred embodiment, the above-mentioned method for dividing the film region and the substrate region is as follows:

[0083] (1) Based on whether the fitted signal corresponding to any three vertex corners is a bimodal signal or a unimodal signal, a label value m is used for labeling. The m value is 1 for bimodal signals and 0 for unimodal signals.

[0084] (2) If the m values ​​of the two vertex corners in the same column are different, the coordinates of the first column pixel point are extracted; if the m values ​​of the two vertex corners in the same row are different, the coordinates of the first row pixel point are extracted; and the range of the dimension coordinate v that changes in the extracted series of pixel point coordinates is determined to be 0 to R-1 or 0 to C-1, where R is the total number of rows in the interference image and C is the total number of columns in the interference image;

[0085] (3) Assign 0 to a, and assign R-1 or C-1 to b, and extract respectively. The series of light intensity contrasts of each pixel are obtained, and after envelope extraction and Gaussian function fitting, the corresponding marker values ​​of the three pixels are obtained. In this case, int() represents integer division;

[0086] (4) If m a and Different values, then With m b If the values ​​are the same, The marker value of each pixel within the range The values ​​are all set to 0. and the current Assign the value of m to b, and repeat step (3); if m a and If they are the same, then With m b Different values ​​will The marker value corresponding to each pixel within the range All values ​​are set to the current m. a The value to be determined, and the current value to be determined. The value of is assigned to 'a', and step (3) is repeated.

[0087] (5) Steps (3) and (4) are executed independently for each other row or column in the area to be detected. For example, if step (2) is to extract the coordinates of the first column of pixels, then steps (3) and (4) are executed independently for each other column. If step (2) is to extract the coordinates of the first row of pixels, then steps (3) and (4) are executed independently for each other row, so as to obtain the m value corresponding to each pixel in the area to be detected, thereby realizing the division of the film area and the substrate area based on the m value.

[0088] The above division method can be simply referred to as the film-substrate dichotomy method. When reconstructing the film thickness in the edge region of the thin film, it is necessary to divide the region into film region and substrate region. To address this, this method proposes the film-substrate dichotomy method, which adaptively identifies and marks the film region and substrate region in the edge region of the thin film. This method can quickly identify and mark the film region and substrate region in each row or column, with low time complexity.

[0089] The value of m corresponding to each pixel can be obtained by a marker matrix based on the pixel coordinates. For regions with a marker matrix of 1 (i.e., thin film region), the number of Gaussian functions is directly set to 2 during iterative fitting; for regions with a marker matrix of 0 (i.e., substrate region), the number of Gaussian functions is directly set to 1 during iterative fitting. Therefore, this method of dividing the thin film region and the substrate region can avoid having to perform two fittings to determine the number of Gaussian functions to identify whether it is a single peak or a double peak when fitting the interference signal of each pixel. This can greatly reduce the algorithm time complexity when reconstructing the morphology of the thin film edge region.

[0090] In summary, this invention proposes a white light interferometric detection method that adapts to multiple regions of printed films, including both the film area and the film edge area, and adapts to the multiple interference signal characteristics of both the film area and the substrate area simultaneously within the field of view when detecting the film edge area under conditions of strong substrate reflectivity. This solves the technical problem that general white light interferometric detection algorithms cannot be universally applied to the detection of multiple regions of printed films, including both the film area and the film edge area, and the detection of multiple interference signal characteristics within a single field of view when detecting the film edge area under conditions of strong substrate reflectivity.

[0091] Example 2

[0092] A novel white light interferometry detection device for displaying thin film morphology in inkjet printing manufacturing includes: a white light scanning interferometry module and a processor; wherein, the white light scanning interferometry module is used to acquire a series of white light scanning interferometry images of the area to be detected in the inkjet printed thin film; the processor executes the novel white light interferometry detection method for displaying thin film morphology in inkjet printing manufacturing as described in Embodiment 1 above.

[0093] The relevant technical solutions are the same as in Embodiment 1, and will not be repeated here.

[0094] Example 3

[0095] A computer-readable storage medium includes a stored computer program, wherein when the computer program is executed by a processor, it controls the device where the storage medium is located to perform a novel white light interferometry detection method for thin film morphology in display inkjet printing manufacturing as described in Embodiment 1 above.

[0096] The relevant technical solutions are the same as in Embodiment 1, and will not be repeated here.

[0097] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A novel white light interferometry method for displaying the film topography in thin film fabrication by inkjet printing, characterized by, The method comprises: acquiring a series of white light scanning interference images of a to-be-detected region of a printed film; extracting a series of light intensity contrasts of any three vertex pixels in the series of images, performing envelope extraction and Gaussian fitting on the series of light intensity contrasts of each vertex, and determining that the to-be-detected region is a film region when the fitting signals of the three vertices are all double-peak signals, and otherwise determining that the to-be-detected region is a film edge region; When it is a thin film region, a series of light intensity contrasts of each pixel point in the region and corresponding series of image acquisition heights are used to construct discrete points for envelope extraction and Gaussian fitting, and the image acquisition height of the double peak in the fitting signal of the corresponding pixel point is determined to calculate the film thickness at the corresponding pixel point; when it is a thin film edge region, the pixel points in the region are divided into film area and substrate area based on envelope extraction and Gaussian fitting, and a series of light intensity contrasts of each pixel point in the film area and corresponding series of image acquisition heights are used to construct discrete points for envelope extraction and Gaussian fitting to determine the image acquisition height of the higher peak in the fitting signal of the corresponding pixel point , a series of light intensity contrasts of a certain pixel point in the substrate area and corresponding series of image acquisition heights are used to construct discrete points for envelope extraction and Gaussian fitting to determine the image acquisition height of the single peak in the fitting signal of the pixel point , the difference between each and is calculated as the film thickness of the corresponding pixel point in the film area, and the detection is completed; wherein the implementation of the film region and the substrate region is: (1) according to whether the fitting signals corresponding to the any three vertexes are double-peak signals or single-peak signals, using a marker value m for marking, the double-peak signals correspond to m value 1, and the single-peak signals correspond to m value 0; (2) if the m values of the two vertexes located in the same column are different, the first column pixel point coordinates are extracted; if the m values of the two vertexes located in the same row are different, the first row pixel point coordinates are extracted; and the dimension coordinate v value range of the extracted series of pixel point coordinates that changes is determined to be 0 to R-1 or 0 to C-1, wherein R is the total number of rows of the interference image, and C is the total number of columns of the interference image; (3) assign 0 to a and R-1 or C-1 to b, respectively extract the series of light intensity contrast of the pixel points at v=a, b, int( ) and obtain the marker values at the three pixel points through the envelope extraction and Gaussian function fitting processing , , , wherein int() represents rounding off; (4) If and Different values, then and If the values ​​are the same, The marker value of each pixel within the range The values ​​are all set to 0. and will the current The value of is assigned to b, and step (3) is repeated; if and If they are the same, then and Different values ​​will The marker value corresponding to each pixel within the range All values ​​are set to the current value. The value to be determined, and the current value to be determined. The value of is assigned to a, and step (3) is repeated. (5) steps (3) and (4) are independently performed on each row or each column in the to-be-detected region to obtain the m values corresponding to each pixel point in the to-be-detected region, and the film region and the substrate region are divided based on the m values.

2. The white light interferometry method of claim 1, wherein, The series of light intensity contrasts of each pixel point are obtained by normalizing and denoising the series of gray values of the pixel point.

3. The white light interferometry method of claim 1, wherein, When the to-be-detected region is a film region, discrete points are constructed using the series of light intensity contrasts of each pixel point in the region and the series of image acquisition relative heights corresponding to the first frame interference image acquisition position, the envelope extraction and Gaussian fitting are performed on the discrete points, and each image acquisition relative height is calculated by multiplying the frame number of the corresponding interference image by the white light interference scanning step distance; the film thickness at each pixel point in the region , the film refractive index, 、 the image acquisition relative height at the double peak, respectively.

4. The white light interferometry method of claim 1, wherein, the envelope extraction is performed on the discrete points constructed based on the series of light intensity contrasts of the target pixel point by using Hilbert transform; the Gaussian fitting is specifically: performing nonlinear iterative fitting on the extracted envelope based on a Gaussian function model to obtain a Gaussian model of the fitting signal of the pixel point.

5. The white light interferometry method of claim 4, wherein, The Gaussian function model is represented as: ； In the formula, is the fitting signal; N represents the number of peaks, and is 0 or 1; is the amplitude corresponding to the i-th Gaussian component; is the center position of the i-th Gaussian component, and is is a Gaussian function expression with a width of . The nonlinear iterative fitting has the following constraints: ， ， ； In the formula, is the weighted error sum of squares of the iteration, is the weight of the i-th envelope discrete point when fitting the envelope; is the value of the light intensity contrast of the i-th envelope discrete point; is the fitted value of the light intensity contrast of the i-th envelope discrete point; is the weighted center height obtained by the barycenter method based on the constructed discrete points; is the image acquisition height at the i-th envelope discrete point; is the adjustment coefficient; is the coherence length of the white light source; is the normalized and denoised preprocessed light intensity contrast of the i-th discrete point constructed.

6. The white light interferometry method of claim 4, wherein, In the envelope extraction and Gaussian fitting of the series of light intensity contrasts of each vertex, or in the division of the film region and the substrate region based on the envelope extraction and Gaussian fitting of the pixel points in the to-be-detected region, the nonlinear iterative fitting is specifically: The number of Gaussian functions in the Gaussian function model is set to 1, and the extracted envelope is subjected to nonlinear iterative fitting to obtain fitting discrete points of the Gaussian function; and based on the fitting discrete points and the discrete points in the extracted envelope, a root mean square error of the iterative fitting is calculated and a determination coefficient ; The number of Gaussian functions in the Gaussian function model is set to 2, and the extracted envelope is subjected to nonlinear iterative fitting, to obtain fitting discrete points after superposition of two Gaussian functions; and based on the fitting discrete points and the discrete points in the extracted envelope, the root mean square error of the iterative fitting is calculated and a determination coefficient ; When and are greater than a preset value, the number N of Gaussian functions is determined according to a formula, otherwise, the nonlinear iterative fitting is re-performed, and finally a fitting signal Gaussian model of the target pixel point is obtained, wherein the formula is: , wherein, when , it is indicated that the fitting signal corresponding to the target pixel point is a single-peak signal; and when , it is indicated that the fitting signal corresponding to the target pixel point is a double-peak signal.

7. The white light interferometry method according to any one of claims 1 to 6, characterized in that Further comprising: using the calculated film thickness of each pixel point in the to-be-detected region to evaluate the film forming quality.

8. A novel white light interferometry setup for displaying the film topography in thin film fabrication, characterized by, The method comprises: a white light scanning interference module and a processor; wherein the white light scanning interference module is used to acquire a series of white light scanning interference images of a to-be-detected region of a printed film; the processor performs a novel white light interference detection method for film topography in display printing manufacturing according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises a stored computer program, wherein the computer program controls the device where the storage medium is located to perform a novel white light interference detection method for film topography in display printing manufacturing according to any one of claims 1 to 7 when the computer program is run by the processor.

Images