Rapid thin film thickness extraction method and device

Abstract

The invention relates to a rapid film thickness extraction method and device, and the method comprises the steps: obtaining the spectrums of a plurality of films, the spectrums comprising a light intensity spectrum, a reflectivity spectrum, a polarization spectrum and a simulation spectrum; determining a hash function of the spectrums, calculating a hash value of each spectrum through the hash function, and mapping a plurality of similar spectrums into a hash bucket through the hash value of each spectrum; constructing a hash index based on each spectrum and the corresponding hash value; acquiring a to-be-measured thin film spectrum and calculating a hash value of the to-be-measured thin film spectrum; on the basis of the Hash value, calculating a Hash bucket where the spectrum of the thin film to be detected is located through a Hash index; and matching the most similar spectrum from the hash bucket, and determining the thickness of the film to be measured according to the most similar spectrum. The spectrum Hash index is constructed through the locality sensitive Hash method, and the spectrum matching efficiency, accuracy and flexibility are improved.

Description

Technical Field

[0001] The invention belongs to the technical field of film thickness measurement and spectroscopy, and in particular relates to a method and a device for quickly extracting film thickness. Background Art

[0002] Accurate measurement of film thickness is of great significance in modern materials science and engineering, especially in the fields of semiconductors, optoelectronic devices, coating materials and nanotechnology. Among the many non-contact measurement techniques, the method based on spectral analysis is widely used due to its high sensitivity and fast response. The Fresnel formula describes the reflection and transmission phenomenon of light at the interface of different media, and can obtain the optical properties of the film by analyzing the phase difference and intensity change between the incident light and the reflected light. In the measurement method based on the Fresnel formula, the optical constants of the film (such as refractive index and extinction coefficient) are usually functions of thickness, so the influence of different thicknesses on the optical constants needs to be considered during measurement. In order to overcome these problems, researchers usually need to establish a relationship model between thickness and spectrum in order to perform effective thickness extraction in actual measurement. The spectral features of films of different thicknesses are calculated in advance and recorded in the spectral library. In actual measurement, the thickness of the film can be quickly and accurately determined by matching the measured spectrum with the spectrum in the spectral library. This method significantly improves the efficiency of measurement and reduces the errors caused by changes in test conditions, thereby ensuring the reliability of the measurement results.

[0003] When it comes to matching algorithms for spectral libraries, the speed issue has always been a bottleneck that needs to be solved urgently. This is mainly reflected in the following aspects: 1. Traditional spectral matching algorithms usually rely on comparing all spectral features one by one. This means that for each spectrum to be measured, it is necessary to perform detailed calculations and comparisons with each sample in the spectral library. This method increases rapidly in the case of large spectral libraries, especially when the number of features reaches tens of thousands, resulting in significantly longer running times. 2. With the advancement of science and technology, the size of spectral libraries has grown exponentially. In modern applications, spectral libraries may contain tens of thousands of spectral data, which makes one-to-one matching even less efficient. In addition, complex measured spectra often contain noise and interference, which further increases the difficulty and calculation time of matching. 3. The one-to-one comparison of traditional matching algorithms not only consumes time, but also takes up a lot of computing resources. Especially when conducting large-scale data analysis, the computer's memory and processing power may become limiting factors. This situation not only affects the efficiency of the matching process, but may also lead to a decline in the overall performance of the system. Summary of the invention

[0004] To improve the efficiency and accuracy of thin film thickness measurement, in the first aspect of the present invention, a fast thin film thickness extraction method is provided, including: obtaining spectra of multiple thin films, where the spectra include light intensity spectra, reflectivity spectra, polarization spectra, and simulation spectra; determining a hash function for the spectra, calculating the hash value of each spectrum through the hash function, and mapping multiple similar spectra into a hash bucket through the hash value of each spectrum; constructing a hash index based on each spectrum and its corresponding hash value; obtaining the spectrum of the thin film to be measured and calculating the hash value of the spectrum of the thin film to be measured; calculating the hash bucket where the spectrum of the thin film to be measured is located through the hash index based on the hash value; matching the most similar spectrum from the hash bucket, and determining the thickness of the thin film to be measured according to the most similar spectrum.

[0005] In some embodiments of the present invention, determining the hash function for the spectra includes: determining one or more hyperplanes based on the spectral data; randomly generating the normal vector and bias term of each hyperplane, and dividing multiple spectral data through the one or more hyperplanes; determining the hash function for all spectral data based on the division result.

[0006] Further, calculating the hash value of each spectrum through the hash function and mapping multiple similar spectra into a hash bucket through the hash value of each spectrum includes: determining the sign function of all hyperplanes based on the division result; calculating the hash value of each spectrum through one or more sign functions; mapping multiple similar spectral data into the hash bucket through the hash value of each spectrum to obtain the hash value of each spectrum.

[0007] In some embodiments of the present invention, constructing a hash index based on each spectrum and its corresponding hash value includes: adding the spectral vector, hash value, and hash bucket index value corresponding to each spectrum into the hash index.

[0008] In some embodiments of the present invention, calculating the hash bucket where the spectrum of the thin film to be measured is located through the hash index based on the hash value includes: calculating the position of the hash bucket where the spectrum of the thin film to be measured is located in the hash index through a preset index function based on the hash value.

[0009] In the above embodiments, matching the most similar spectrum from the hash bucket includes: extracting one or more candidate spectral vectors from the hash bucket, calculating the distance between each candidate spectral vector and the spectral vector of the spectrum of the thin film to be measured; taking the spectrum corresponding to the candidate spectral vector with the minimum distance as the most similar spectrum.

[0010] In a second aspect of the present invention, there is provided a fast thin-film thickness extraction device, comprising: an acquisition module for acquiring spectra of a plurality of thin films, the spectra including light intensity spectra, reflectance spectra, polarization spectra, and simulation spectra; a construction module for determining a hash function of the spectra, calculating a hash value of each spectrum through the hash function, and mapping a plurality of similar spectra to a hash bucket through the hash value of each spectrum; constructing a hash index based on each spectrum and its corresponding hash value; a calculation module for acquiring a spectrum of a thin film to be measured and calculating a hash value of the spectrum of the thin film to be measured; calculating, based on the hash value, the hash bucket where the spectrum of the thin film to be measured is located through the hash index; a matching module for matching the most similar spectrum from the hash bucket and determining the thickness of the thin film to be measured according to the most similar spectrum.

[0011] In a third aspect of the present invention, there is provided an electronic device, comprising: one or more processors; a storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the fast thin-film thickness extraction method provided by the present invention in the first aspect.

[0012] In a fourth aspect of the present invention, there is provided a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the fast thin-film thickness extraction method provided by the present invention in the first aspect.

[0013] The beneficial effects of the present invention are as follows: By mapping data to hash buckets, LSH can significantly reduce the number of candidates to be compared in approximate nearest neighbor search, thereby significantly reducing the query time complexity. The query time is usually O(k), where k is the number of neighbors to be returned. Brute-force search requires calculating the distance for each point in the dataset, with a time complexity of O(n), where n is the number of data points, which is very inefficient in large-scale datasets.

[0014] On the other hand, LSH has excellent dynamic update capabilities and can quickly support addition and deletion operations on online datasets. Simply map the new data to the hash bucket or directly remove the corresponding data without having to recalculate the distances of all data points. This makes LSH particularly suitable for tasks requiring real-time feedback. In contrast, brute-force search requires full recalculation when data is updated, resulting in low efficiency and inability to meet real-time requirements. Therefore, LSH shows significant advantages in the face of dynamic data scenarios. Description of the Drawings

[0015] Figure 1 It is a schematic diagram of the basic process of the fast thin-film thickness extraction method in some embodiments of the present invention; Figure 2Schematic flow chart of the fast film thickness extraction method in some embodiments of the present invention; Figure 3 Schematic diagram of the principle of spectral dimensionality reduction using two hyperplanes through a hash function in some embodiments of the present invention; Figure 4 Schematic diagram of the principle of mapping similar spectra to the same bucket and encoding them by two hyperplanes in some embodiments of the present invention; Figure 5 Schematic structural diagram of the fast film thickness extraction device in some embodiments of the present invention; Figure 6 Schematic structural diagram of an electronic device in some embodiments of the present invention. Detailed implementation manners

[0016] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not intended to limit the scope of the present invention.

[0017] Refer to Figure 1 and Figure 2 , in the first aspect of the present invention, a fast film thickness extraction method is provided, including: S100. Obtain spectra of multiple films, where the spectra include intensity spectra, reflectivity spectra, polarization spectra, and simulation spectra; S200. Determine the hash function of the spectra, calculate the hash value of each spectrum through the hash function, and map multiple similar spectra to a hash bucket through the hash value of each spectrum; based on each spectrum and the corresponding hash value, construct a hash index; S300. Obtain the spectrum of the film to be measured and calculate the hash value of the spectrum of the film to be measured; based on the hash value, calculate the hash bucket where the spectrum of the film to be measured is located through the hash index; S400. Match the most similar spectrum from the hash bucket, and determine the thickness of the film to be measured according to the most similar spectrum.

[0018] It can be understood that the spectra in this patent include but are not limited to Fresnel modeled reflectivity spectra, other intensity spectra, reflectivity spectra, ellipsometry spectra, etc. of the wavelength distribution measured actually, or simulation reflectivity spectra, ellipsometry spectra, etc. calculated by the thin film transfer matrix. Such spectral forms also belong to the protection scope of this application. And Fresnel reflectivity is a key parameter describing the reflection and transmission characteristics of light at the interface of a medium. In a multi-layer film structure, the reflection and transmission of light not only depend on the incident angle and wavelength of the incident light, but also are affected by the refractive index and thickness of each layer of the film. Reflectivity refers to the proportion of the light intensity incident on the medium interface that is reflected back. For any two media with different refractive indices, the reflectivity can be calculated by the Fresnel equation.

[0019] In step S100 of some embodiments of the present invention, spectra of multiple thin films are obtained, and the spectra include light intensity spectra, reflectivity spectra, polarization spectra, and simulation spectra; S101. Assume that light is incident from medium 1 (refractive index ) to medium 2 (refractive index ), then the calculation of reflectivity can be divided into s-polarized light and p-polarized light: Reflectivity of s-polarized light: , Reflectivity of p-polarized light: , where and are the refractive indices of the media, is the angle of incidence. is the angle of refraction, which can be calculated by Snell's law: ; In the case of a multi-layer film, when light waves pass through each layer of the film, reflection and transmission will occur. The thickness and refractive index of each layer will affect the phase and intensity of the light waves. To calculate the total reflectivity, a recursive method needs to be used to calculate the reflectivity of each layer from the last layer forward.

[0020] S102. When light waves pass through the film layer, their propagation in the film causes a change in phase. The phase change can be expressed as:

[0021] where λ is the wavelength of light, is the thickness of the j-th layer, is the refractive index of this layer is the angle of refraction of the light after entering this layer of the film.

[0022] S103. Recursively calculate the total reflectivity of the material In the calculation of a multi-layer film, calculate the reflectivity recursively from the last layer to the first layer: Initialize the reflection coefficient: Initialize the reflection coefficient between the bottom layer (substrate) and the top layer (air).

[0023] Calculate from the last layer forward: For each layer of the film, calculate its reflectivity to the previous layer using the following formula: , , where and are the reflectivities of the previous layer of the film, and are the reflectivities of the current layer of the film.

[0024] S104. Calculation of total reflectance: The final total reflectance can be calculated from s-polarization and p-polarization as follows: .

[0025] It can be understood that LSH is a hashing technique that makes it more likely for similar data to collide (i.e., map to the same bucket) in a hash table. Its core idea is to design a hashing function such that similar input points result in similar hash values after passing through the hashing function. Among them, the random partitioning hyperplane is a key technique in Locality-Sensitive Hashing (LSH). It divides the high-dimensional space into multiple parts by randomly generating hyperplanes, so that similar data points are more likely to be mapped to the same hash bucket in the same part.

[0026] Reference Figure 3 And Figure 4 , in step S200 of some embodiments of the present invention, the determining the hashing function of the spectrum includes: S201. Determine one or more hyperplanes based on the spectral data; Specifically, in a d-dimensional space, a hyperplane can be represented by a linear equation: , where w is the normal vector of the hyperplane, R represents the spectral quantity that varies with wavelength, specifically including the light intensity spectrum, reflectance spectrum, or other forms of spectra, etc. Among them, the light intensity spectrum is the radiation intensity distribution of the actual measured electromagnetic wave at different wavelengths; the reflectance spectrum describes the change in the reflectance of the sample surface to incident light at different wavelengths, which can be obtained through experimental measurement or simulated calculation based on the Fresnel formula or thin film transfer matrix method; other spectra include but are not limited to spectra of polarization state change amounts, etc. R In different measurement or calculation scenarios, the specific form representing the above spectra b is the bias term.

[0027] S202. Randomly generate the normal vector and bias term of each hyperplane, and divide multiple spectral data by one or more hyperplanes; Specifically, to ensure consistent results each time, a random seed is used to generate fixed random numbers: , where represents generating d d-dimensional random vectors from a normal distribution with a mean of 0 and a variance of 1.

[0028] Bias term bIt can be randomly generated from a certain distribution. A common approach is to select a random value, usually from a uniform distribution: , where the m,n depends on the range of the data.

[0029] S203. Based on the segmentation result, determine the hash function for all spectral data.

[0030] Specifically, given a spectrum R , we can determine on which side of the hyperplane it lies by calculating the value of the hyperplane equation: If , then R it is on the positive side of the hyperplane; If , then R it is on the negative side of the hyperplane.

[0031] Furthermore, in step S204, the calculating the hash value of each spectrum through the hash function and mapping multiple similar spectra to a hash bucket through the hash value of each spectrum includes: determining the sign function of all hyperplanes based on the segmentation result; calculating the hash value of each spectrum through one or more sign functions; mapping multiple similar spectral data to the hash bucket through the hash value of each spectrum to obtain the hash value of each spectrum.

[0032] Specifically, using these hyperplanes, a hash function can be constructed, where k is the number of hyperplanes and is a constant. Each hyperplane generates a binary bit for the input point, indicating the position of the point: ; Here, sgn is the sign function, which limits the output to 0 or 1, so that the data points can be mapped to the hash space. Finally, LSH will use multiple hash functions R for the given spectral vector to generate a hash value, where k is a constant. This hash value can be regarded as a binary string, indicating the position of the input vector in the hash space.

[0033] In step S205 of some embodiments of the present invention, the constructing the hash index based on each spectrum and the corresponding hash value includes: adding the spectral vector, hash value, and hash bucket index value corresponding to each spectrum to the hash index.

[0034] In step S300 of some embodiments of the present invention, calculating the hash bucket where the measured thin-film spectrum is located based on the hash value includes: calculating the position of the hash bucket where the measured thin-film spectrum is located in the hash index based on the hash value through a preset index function.

[0035] Specifically, each vector in all simulation spectrum vectors is added to the Locality-Sensitive Hashing (LSH) index. The LSH index uses a locality-sensitive hashing function to map the input vector into a hash bucket. In this way, similar vectors have a higher probability of being mapped to the same bucket, so that potential similar items can be quickly retrieved in subsequent similarity queries. During the indexing process, the original spectrum vector data will be stored together with its corresponding hash value to ensure convenient access to this data when needed.

[0036] Mapping the hash value to the hash bucket: ; Storing the spectrum vector and its corresponding hash value into the LSH index: , where R j represents the j th spectrum in the library, B represents the total number of hash buckets.

[0037] In this way, all vectors are effectively added to the LSH index to form a structured data storage. This process not only improves the efficiency of subsequent similarity queries but also lays a foundation for subsequent data processing and analysis. Moreover, after using the index, users can quickly find data points similar to a specific query vector, significantly accelerating the similarity search process.

[0038] It should be noted that under the bucket-based structure, LSH will look for buckets with the same or similar hash values as the measured vector and extract possible similar vectors from these buckets. This step greatly reduces the number of vectors that need to be compared, thereby improving the query efficiency. For the candidate vectors extracted from the similar buckets, a specified distance function (the distance functions in this patent include but are not limited to methods such as Euclidean distance, cosine similarity, and L1 norm) is used to calculate the distance between these candidate vectors and the measured spectrum vector. This is a one-by-one comparison process. According to the distance value, the similarity is judged, and the simulation spectrum with the smallest distance and its index in the library are returned according to the distance size. Finally, the thickness is extracted according to the index.

[0039] Therefore, in step S400 of the above embodiments, matching the most similar spectrum from the hash bucket includes: S401. Extract one or more candidate spectral vectors from the hash bucket, and calculate the distance between each candidate spectral vector and the spectral vector of the thin film to be measured; Specifically, calculate the Euclidean distance: , where, R i represents the i th element of the spectral vector in the library, r i represents the i th element of the measured spectral vector, N represents the number of spectral elements, d represents the Euclidean distance between the measured spectrum and the spectrum in the library. The smaller this value is, the closer the measured spectrum and the spectrum in the library are.

[0040] Optionally, use the Euclidean distance, chi-square distance, geodesic distance or other distance metric methods to calculate the distance between the candidate spectral vector and the spectral vector of the thin film to be measured.

[0041] S402. Take the spectrum corresponding to the candidate spectral vector with the smallest distance as the most similar spectrum.

[0042] Compared with the traditional exhaustive method, according to the SOI (Si + SiO2) structure, spectral libraries with different spectral density sizes were generated for testing. One point represents one spectrum. The comparison results are shown in the following table:

[0043] It can be seen from the table that as the library scale increases, the time-consuming of spectral matching will become longer and longer. Using the exhaustive method can no longer meet the requirements of fast real-time measurement, while using the LSH algorithm can effectively improve the spectral matching efficiency.

[0044] Embodiment 2 Reference Figure 5 , a second aspect of the present invention provides a fast thin film thickness extraction device 1, including: an acquisition module 11 for acquiring the spectra of multiple thin films, the spectra including light intensity spectra, reflectivity spectra, polarization spectra and simulation spectra; a construction module 12 for determining the hash function of the spectra, calculating the hash value of each spectrum through the hash function, and mapping multiple similar spectra to a hash bucket through the hash value of each spectrum; constructing a hash index based on each spectrum and the corresponding hash value; a calculation module 13 for acquiring the spectrum of the thin film to be measured and calculating the hash value of the spectrum of the thin film to be measured; calculating the hash bucket where the spectrum of the thin film to be measured is located through the hash index based on the hash value; a matching module 14 for matching the most similar spectrum from the hash bucket and determining the thickness of the thin film to be measured according to the most similar spectrum.

[0045] The building block 12 includes: a first determination unit configured to determine one or more hyperplanes based on the spectral data; a generation unit configured to randomly generate a normal vector and a bias term for each hyperplane, and segment a plurality of spectral data by the one or more hyperplanes; and a second determination unit configured to determine a hash function for all the spectral data based on the segmentation result.

[0046] Embodiment 3 Reference Figure 6 In a third aspect of the present invention, there is provided an electronic device, including: one or more processors; a storage device configured to store one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the fast thin film thickness extraction method of the first aspect of the present invention.

[0047] The electronic device 500 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 501, which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 are also stored. The processing device 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0048] Generally, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 500 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 6 the electronic device 500 with various devices is shown, it should be understood that it is not required to implement or include all the shown devices. Instead, more or fewer devices may be implemented or included. Figure 6 Each block shown in

[0049] In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present disclosure includes a computer program product that includes a computer program carried on a computer-readable medium, and the computer program includes program code for performing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by a processing device 501, the above-described functions defined in the methods of the embodiments of the present disclosure are performed. It should be noted that the computer-readable medium described in the embodiments of the present disclosure can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the embodiments of the present disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the embodiments of the present disclosure, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and the computer-readable signal medium can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0050] The above computer-readable medium can be included in the above electronic device; or it can exist separately without being assembled into the electronic device. The above computer-readable medium carries one or more computer programs, and when the above one or more programs are executed by the electronic device, the electronic device is caused to: Computer program code for performing the operations of the embodiments of the present disclosure may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, Python, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0051] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should be noted that each block in the block diagram and / or flowchart, and the combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system that performs the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0052] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for rapid film thickness extraction, characterized in that: include: Acquire spectra of multiple thin films, wherein the spectra include a light intensity spectrum, a reflectivity spectrum, a polarization spectrum, and a simulation spectrum; Determine a hash function of the spectrum, calculate a hash value of each spectrum by the hash function, and map multiple similar spectra into a hash bucket by the hash value of each spectrum; Based on each spectrum and the corresponding hash value, a hash index is constructed; Obtaining a spectrum of the film to be tested and calculating a hash value of the spectrum of the film to be tested; Based on the hash value, the hash bucket where the spectrum of the thin film to be tested is located is calculated by hash index; The most similar spectrum is matched from the hash bucket, and the thickness of the film to be measured is determined according to the most similar spectrum.

2. The rapid film thickness extraction method according to claim 1, characterized in that: The determining of the hash function of the spectrum comprises: determining one or more hyperplanes based on the spectral data; The normal vector and bias term of each hyperplane are randomly generated, and multiple spectral data are segmented by one or more hyperplanes; Based on the segmentation results, a hash function for all spectral data is determined.

3. The rapid film thickness extraction method according to claim 2, characterized in that: The step of calculating the hash value of each spectrum by a hash function and mapping multiple similar spectra into a hash bucket by the hash value of each spectrum includes: Determine the sign functions of all hyperplanes based on the segmentation results; calculate the hash value of each spectrum through one or more sign functions; Multiple similar spectral data are mapped into hash buckets through the hash value of each spectrum to obtain the hash value of each spectrum.

4. The rapid film thickness extraction method according to claim 1, characterized in that: The constructing of a hash index based on each spectrum and the corresponding hash value includes: The spectral vector, hash value, and hash bucket index value corresponding to each spectrum are added to the hash index.

5. The rapid film thickness extraction method according to claim 1, characterized in that: The step of calculating the hash bucket where the spectrum of the thin film to be tested is located by hash index based on the hash value includes: Based on the hash value, the position of the hash bucket where the spectrum of the thin film to be measured is located in the hash index is calculated through a preset index function.

6. The rapid film thickness extraction method according to claim 1, characterized in that: The matching the most similar spectrum from the hash bucket comprises: Extract one or more candidate spectral vectors from the hash bucket, and calculate the distance between each candidate spectral vector and the spectral vector of the thin film spectrum to be measured; The spectrum corresponding to the candidate spectrum vector with the smallest distance is taken as the most similar spectrum.

7. A rapid film thickness extraction device, characterized in that: include: An acquisition module, used for acquiring spectra of multiple thin films, wherein the spectra include a light intensity spectrum, a reflectivity spectrum, a polarization spectrum and a simulation spectrum; A construction module is used to determine a hash function of the spectrum, calculate a hash value of each spectrum by the hash function, and map multiple similar spectra into a hash bucket by the hash value of each spectrum; Based on each spectrum and the corresponding hash value, a hash index is constructed; A calculation module, used for obtaining the spectrum of the thin film to be tested and calculating the hash value of the spectrum of the thin film to be tested; Based on the hash value, the hash bucket where the spectrum of the thin film to be tested is located is calculated by hash index; The matching module is used to match the most similar spectrum from the hash bucket and determine the thickness of the film to be measured according to the most similar spectrum.

8. The rapid film thickness extraction device according to claim 7, characterized in that: The building blocks include: A first determining unit, configured to determine one or more hyperplanes based on the spectral data; A generating unit, used for randomly generating a normal vector and a bias item of each hyperplane, and segmenting a plurality of spectral data by one or more hyperplanes; The second determining unit is used to determine the hash function of all the spectral data based on the segmentation result.

9. An electronic device, comprising: one or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the fast film thickness extraction method according to any one of claims 1 to 6.

10. A computer readable medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the fast film thickness extraction method according to any one of claims 1 to 6 is implemented.

Images