WO2019127352A1 - 基于拉曼光谱的物质识别方法及云端*** - Google Patents
基于拉曼光谱的物质识别方法及云端*** Download PDFInfo
- Publication number
- WO2019127352A1 WO2019127352A1 PCT/CN2017/119796 CN2017119796W WO2019127352A1 WO 2019127352 A1 WO2019127352 A1 WO 2019127352A1 CN 2017119796 W CN2017119796 W CN 2017119796W WO 2019127352 A1 WO2019127352 A1 WO 2019127352A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- substance
- detected
- component
- task
- raman
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
- G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
- G01N21/65—Raman scattering
Definitions
- the present application relates to the field of material recognition technology, and in particular to a material recognition method based on Raman spectroscopy and a cloud system.
- the Raman spectrum is a spectrum in which monochromatic light passes through a transparent medium and the frequency of light scattered by the molecules changes.
- the Raman spectrum reflects the vibration characteristics of the molecule and can be used to detect the substance. That is, the Raman spectrum recognition technology can identify the substance composition according to the Raman spectrum formed by the substance to be detected.
- Raman spectroscopy technology can qualitatively analyze materials in a simple, rapid and non-destructive manner. There is no special requirement for the environment and no need to treat the substances to be tested, which reduces the errors caused by the processing of the substances themselves. With the rapid development of devices such as lasers, more and more miniaturized, intelligent, and inexpensive Raman spectroscopy equipment has entered the market.
- the problem in the prior art is that the Raman spectral recognition technology used in the Raman spectroscopy detection apparatus easily recognizes the noise peak of the Raman spectrum as the characteristic peak of the Raman spectrum when extracting the characteristic peak of the Raman spectrum, and needs By identifying the material components in a way that is compared with the sample library, the recognition speed decreases as the database expands; and the security of the database on the Raman spectrum detection device side is poor, and the encryption protection of the database also has certain limitations.
- the embodiment of the present application proposes a material recognition method based on Raman spectroscopy and a cloud system to solve the problem that the existing Raman detection device has poor recognition accuracy, low recognition speed and Raman detection device side when performing material recognition. Less secure technical issues.
- an embodiment of the present application provides a material recognition method based on Raman spectroscopy, including:
- the Raman spectrum of the substance to be detected is identified based on a substance recognition model of the preset multi-task learning, and the substance composition of the substance to be detected and the ratio thereof are obtained.
- an embodiment of the present application provides a Raman spectroscopy-based substance recognition cloud system, including:
- Raman spectroscopy acquisition terminal for collecting Raman spectroscopy data of the substance to be detected
- a recognition server for receiving Raman spectral data of the substance to be detected
- the Raman spectrum of the substance to be detected is identified based on a substance recognition model of the preset multi-task learning, and the substance composition of the substance to be detected and the ratio thereof are obtained.
- an embodiment of the present application provides an electronic device, where the electronic device includes:
- Receiving device memory, one or more processors
- One or more modules the one or more modules being stored in the memory and configured to be executed by the one or more processors, the one or more modules comprising Instructions for each step.
- embodiments of the present application provide a computer program product for use with an electronic device, the computer program product comprising a computer readable storage medium and a computer program mechanism embedded therein, the computer program mechanism Instructions are included for performing the various steps in the above methods.
- the Raman spectrum data of the substance to be detected sent from the Raman spectrum detecting device is received, and the Raman spectrum of the substance to be detected is identified based on a predetermined substance recognition model of multi-task learning.
- the substance composition of the detected substance and its proportion are described and sent to the terminal for display.
- FIG. 1 is a schematic diagram of a material recognition method based on Raman spectroscopy in the first embodiment of the present application
- FIG. 2 is a schematic flow chart of a method for identifying a substance based on Raman spectroscopy in the first embodiment of the present application
- FIG. 3 is a structural diagram of a material recognition cloud system based on Raman spectroscopy in Embodiment 2 of the present application;
- FIG. 4 is a schematic structural diagram of an electronic device according to Embodiment 3 of the present application.
- the traditional Raman spectral recognition technology can easily identify the noise peak of the Raman spectrum as the characteristic peak of the Raman spectrum when extracting the characteristic peak of the Raman spectrum, and it is necessary to identify the material composition by comparing with the sample library.
- the expansion of the database is reduced; and the security of the on-board database is weak, and even if the database is modified, effective encryption protection cannot be achieved.
- the embodiment of the present application proposes a deep learning algorithm based on convolutional neural network, learns and extracts the feature vector of the Raman spectrum, directly processes and recognizes the Raman spectrum formed by the detected substance, and realizes the substance.
- the identification of the composition and its proportion compared with the traditional Raman spectroscopy identification technology, does not need to extract the characteristic peak of the Raman spectrum of the substance to be detected, so that it is not susceptible to noise interference in the Raman spectrum, and will not be due to the expansion of the database. Affect the recognition speed.
- the database of the Raman spectroscopy detection device is deployed in the cloud, and the database is not set on the Raman spectroscopy detection device side, which reduces the cost of the Raman spectroscopy detection device while ensuring the security of the database.
- the embodiments of the present application are based on a network, cloud computing, deep learning and other technologies, that is, a material recognition model based on cloud-based multi-task learning, and a Raman spectrum of a substance to be detected is identified by a substance composition and a substance composition ratio, wherein the substance composition is identified and Material composition ratio identification Two identification tasks are separated in the same substance recognition model, so as to achieve the technical effect of simple identification architecture, high model reuse rate and low recognition speed affected by database expansion.
- FIG. 1 is a schematic diagram showing a method for identifying a substance based on Raman spectroscopy in the first embodiment of the present application
- FIG. 2 is a schematic flow chart showing a method for identifying a substance based on Raman spectroscopy in the first embodiment of the present application, as shown in FIG. 1 and FIG. As shown, the method includes:
- Step 101 Receive Raman spectrum data of the substance to be detected.
- Step 102 Identify a Raman spectrum of the substance to be detected based on a substance recognition model of the preset multi-task learning, and obtain a material composition of the substance to be detected and a ratio thereof.
- a Raman spectroscopy acquisition terminal ie, a Raman spectroscopy detection device
- measures the substance to be detected to obtain Raman spectroscopy data and transmits the Raman spectroscopy data to the data transmission module of the Raman spectroscopy acquisition terminal via the network to
- the identification server of the cloud system the identification server receives Raman spectral data of the substance to be detected.
- the identification server identifies the received Raman spectral data of the substance to be detected, and transmits the obtained recognition result of the substance component of the substance to be detected and the ratio thereof to the Raman spectrum collecting terminal and displays the recognition result.
- the establishment of the preset material recognition model for multi-task learning includes:
- the initialized material recognition model is trained to obtain the material recognition model of the trained multi-task learning.
- the training server trains the material recognition model of the multi-task learning based on the initial Raman spectral data in the database, and deploys the trained multi-task learning material recognition model to the identification server without deploying the database to the identification server.
- the specific implementation method of the training server to train the multi-task learning material recognition model based on the initial Raman spectral data in the database is:
- the Raman spectral data of all the single substances in the current database are arranged in pairs, and the two-two combinations of Raman spectral data constituting all the substances, for example, the combination of the Raman spectral data of the substance A and the substance B,
- the ratio of the components is proportional to the ratio of the components, starting from 0, increasing to 1 in steps of 0.05, and combining 20 different components, that is, substance A and substance B are 0% and 100%, 5% and 95. %, 10%, and 90%, etc., and so on until the composition ratio is 0% and 100%.
- the new Raman spectral data is used as training.
- the sample trains the initialized material recognition model and obtains the material recognition model of the trained multi-task learning to realize the identification of the material composition of the substance to be detected and its proportion.
- the multitasking in the material recognition model of the preset multitasking learning includes a first task for material component identification and a second task for material component ratio recognition.
- the calculation formula of the loss function of the material recognition model of the preset multi-task learning is:
- Loss function 0.5 * material component loss function + 0.5 * material component proportional loss function.
- the optimization goal of the first task in the material recognition model of multi-task learning is the material component
- the optimization goal of the second task is the material component ratio.
- the Raman spectrum of the substance to be detected is identified, and the material composition of the substance to be detected and the ratio thereof are obtained, including:
- the ratio of each of the substance components is identified based on the second task based on the respective substance component numbers.
- the Raman spectral data of the substance to be detected is used as an input of a substance recognition model for identifying multi-task learning in the server, and the feature vector of the Raman spectral data is extracted, and each of the substances to be detected may be included based on the first task identification.
- the material component number and the confidence level corresponding to each substance component if the confidence level corresponding to each substance component reaches the confidence threshold of each substance component set in the first task, for example, includes two substance component numbers W 0 and W 1 ,
- the confidence levels correspond to P 0 and P 1 respectively
- the confidence thresholds corresponding to the two substance components are P t0 and P t1
- the substance composition numbers of the substances to be detected are determined to be W 0 and W 1 , based on the second task pair.
- the determined ratio of the substance components of the substance to be detected is identified; if the confidence levels corresponding to the substance components do not all reach the confidence threshold of each substance component set in the first task, for example, three substance component numbers W 0 , W are included.
- the method further includes:
- the information of each substance component is obtained based on the component number of each substance contained in the substance to be detected.
- the database in order to ensure that the Raman spectral data in the database is not leaked, the database is deployed in the training server, and the database configures the material composition information and the corresponding material component number for all Raman spectral data, so that the database numbers and corresponding the material components.
- the substance component information is transmitted to the identification server, and after the identification server recognizes the substance component numbers included in the substance to be detected, the database is accessed based on each substance component number, and the substance component information corresponding to each substance component number in the database is acquired.
- Embodiment 1 of the present application provides a detailed description of Embodiment 1 of the present application by taking a specific scenario as an example.
- the application range of the embodiments of the present application includes, but is not limited to, Raman spectroscopy-based mixture quality identification, and Raman spectroscopy-based mixture identification is taken as an example.
- the Raman spectroscopy-based substance recognition cloud system includes a Raman spectroscopy acquisition terminal, a recognition server, and Train the server, the specific process is as follows:
- Step 201 Combine the Raman spectral data of all the single substances in the database into different composition ratios to form new Raman spectral data.
- Step 202 Using the training server, training the initialized material recognition model with the new Raman spectral data as a training sample, and obtaining a trained material recognition model for multi-task learning, which is used for the substance composition and the proportion of the substance to be detected. Identification.
- Step 203 Deploy the material identification model of the trained multi-task learning in the identification server.
- Step 204 The Raman spectrum acquisition terminal measures the detected substance to obtain Raman spectrum data, and performs Raman spectrum data calibration, de-base noise processing, and the like, and uploads it to the identification server of the cloud system.
- Step 205 The identification server performs normalization processing on the received Raman spectral data of the substance to be detected, and the normalization process is specifically: normalizing the range and resolution of the x-axis wave number of the Raman spectral data. .
- Step 206 Input the normalized Raman spectral data of the substance to be detected into the material recognition model of the multi-task learning, extract the feature vector of the Raman spectral data, and output the possible substances to be detected according to the first task.
- the material component number and the confidence level corresponding to each substance component determine the component number of each substance contained in the substance to be detected based on the confidence threshold corresponding to each substance component, and the ratio of each substance component in the substance to be detected based on the second task.
- Step 207 Access the database according to each component number of the substance to be detected, and acquire the substance component information corresponding to each component number in the database.
- Step 208 Send the substance composition information and the substance composition ratio of the substance to be detected to the Raman spectrum collection terminal, so that the Raman spectrum collection terminal displays the substance composition information and the substance composition ratio of the detection substance.
- a Raman spectroscopy-based substance recognition cloud system is also provided in the embodiment of the present application. Since the principle of solving the problem of these devices is similar to a material recognition method based on Raman spectroscopy, the implementation of these devices is implemented. See the implementation of the method, and the repetition will not be repeated.
- FIG. 3 is a structural diagram of a material recognition cloud system based on Raman spectroscopy in Embodiment 2 of the present application.
- the Raman spectroscopy-based substance recognition cloud system 300 may include:
- the Raman spectrum acquisition terminal 301 is configured to collect Raman spectrum data of the substance to be detected.
- An identification server 302 configured to receive Raman spectral data of the substance to be detected
- the Raman spectrum of the substance to be detected is identified based on a substance recognition model of the preset multi-task learning, and the substance composition of the substance to be detected and the ratio thereof are obtained.
- the training server 303 is configured to establish the substance identification model of the preset multi-task learning, and the training server 303 includes:
- the initialized material recognition model is trained to obtain the material recognition model of the trained multi-task learning.
- the multitasking in the material recognition model of the preset multitasking learning includes a first task for material component identification and a second task for material component ratio recognition.
- the calculation formula of the loss function of the material recognition model of the preset multi-task learning is:
- Loss function 0.5 * material component loss function + 0.5 * material component proportional loss function.
- the Raman spectrum of the substance to be detected is identified, and the material composition of the substance to be detected and the ratio thereof are obtained, including:
- the ratio of each of the substance components is identified based on the second task based on the respective substance component numbers.
- the method further includes:
- the information of each substance component is obtained based on the component number of each substance contained in the substance to be detected.
- an electronic device is also provided in the embodiment of the present application. Since the principle is similar to a material recognition method based on Raman spectroscopy, the implementation of the method can be referred to the implementation of the method, and the repeated description is not repeated.
- the electronic device includes: a transceiver device 401, a memory 402, one or more processors 403, and one or more modules.
- the one or more modules are stored in the memory and configured to be executed by the one or more processors, the one or more modules including steps for performing the steps of any of the above methods instruction.
- the embodiment of the present application further provides a computer program product for use in combination with an electronic device. Since the principle is similar to a material recognition method based on Raman spectroscopy, the implementation can be referred to the implementation of the method. It will not be repeated here.
- the computer program product comprises a computer readable storage medium and a computer program mechanism embodied therein, the computer program mechanism comprising instructions for performing the various steps of any of the above methods.
- embodiments of the present application can be provided as a method, system, or computer program product.
- the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment in combination of software and hardware.
- the application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
- the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
- the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
- These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
- the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.
Landscapes
- Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
Abstract
基于拉曼光谱的物质识别方法及云端***(300)，方法包括：接收待检测物质的拉曼光谱数据(101)；基于预设的多任务学习的物质识别模型，对待检测物质的拉曼光谱进行识别，得到待检测物质的物质成分及其比例(102)。相较于传统拉曼光谱识别技术，该方法不需要提取待检测物质的拉曼光谱的特征峰，从而不易受到拉曼光谱中噪声的干扰，以及不会因为数据库的扩大而影响识别速度，达到加速识别速度的技术效果。
Description
本申请涉及物质识别技术领域，特别涉及基于拉曼光谱的物质识别方法及云端***。
拉曼光谱是单色光穿过透明介质，被分子散射的光发生频率变化的光谱。拉曼光谱反映的是分子的振动特征，可用于对物质的检测，即拉曼光谱识别技术能够根据待检测物质形成的拉曼光谱识别物质成分。
拉曼光谱识别技术能够简单、快速、无损地对物质进行定性分析，没有对环境的特殊要求，以及不需要对待检测物质进行处理，减少了由于对物质本身进行处理带来的误差，因此，随着激光器等设备的快速发展，越来越多的小型化、智能化、价格低廉的拉曼光谱检测设备进入市场。
现有技术存在的问题是，拉曼光谱检测设备所采用的拉曼光谱识别技术在提取拉曼光谱的特征峰时容易将拉曼光谱的噪声峰误识别为拉曼光谱的特征峰，而且需要通过与样本库比对的方式识别物质成分，识别速度随着数据库的扩大而降低；以及拉曼光谱检测设备侧的数据库的安全性较差，同时对数据库进行加密保护也存在一定的局限性。
发明内容
本申请实施例提出了基于拉曼光谱的物质识别方法及云端***，以解决现有拉曼检测设备在进行物质识别时，识别精确度较差、识别速度较低以及拉曼检测设备侧的数据库安全性较低的技术问题。
在一个方面，本申请实施例提供了一种基于拉曼光谱的物质识别方法， 包括：
接收待检测物质的拉曼光谱数据；
基于预设的多任务学习的物质识别模型，对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例。
在另一个方面，本申请实施例提供了一种基于拉曼光谱的物质识别云端***，包括：
拉曼光谱采集终端，用于采集待检测物质的拉曼光谱数据；
识别服务器，用于接收待检测物质的拉曼光谱数据；以及，
基于预设的多任务学习的物质识别模型，对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例。
在另一个方面，本申请实施例提供了一种电子设备，所述电子设备包括：
接收设备，存储器，一个或多个处理器；以及
一个或多个模块，所述一个或多个模块被存储在所述存储器中，并被配置成由所述一个或多个处理器执行，所述一个或多个模块包括用于执行上述方法中各个步骤的指令。
在另一个方面，本申请实施例提供了一种与电子设备结合使用的计算机程序产品，所述计算机程序产品包括计算机可读的存储介质和内嵌于其中的计算机程序机制，所述计算机程序机制包括用于执行上述方法中各个步骤的指令。
有益效果如下：
本实施例中，接收来自拉曼光谱检测设备发送的待检测物质的拉曼光谱数据，基于预设的多任务学习的物质识别模型，对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例，并发送给终端进行显示。相较于传统拉曼光谱识别技术，不需要提取待检测物质的拉 曼光谱的特征峰，从而不易受到拉曼光谱中噪声的干扰，以及不会因为数据库的扩大而影响识别速度，达到加速识别速度的技术效果。
下面将参照附图描述本申请的具体实施例，其中：
图1为本申请实施例一中基于拉曼光谱的物质识别方法原理图；
图2为本申请实施例一中基于拉曼光谱的物质识别方法流程示意图；
图3为本申请实施例二中基于拉曼光谱的物质识别云端***架构图；
图4为本申请实施例三中电子设备的结构示意图。
以下通过具体示例，进一步阐明本发明实施例技术方案的实质。
为了使本申请的技术方案及优点更加清楚明白，以下结合附图对本申请的示例性实施例进行进一步详细的说明，显然，所描述的实施例仅是本申请的一部分实施例，而不是所有实施例的穷举。并且在不冲突的情况下，本说明中的实施例及实施例中的特征可以互相结合。
发明人在发明过程中注意到：
传统拉曼光谱识别技术在提取拉曼光谱的特征峰时容易将拉曼光谱的噪声峰误识别为拉曼光谱的特征峰，而且需要通过与样本库比对的方式识别物质成分，识别速度随着数据库的扩大而降低；以及机载数据库安全性的保障能力较弱，即便对数据库进行改造仍无法实现有效的加密保护。
针对上述不足/基于此，本申请实施例提出了基于卷积神经网络的深度学习算法，学习并提取拉曼光谱的特征向量，直接对待检测物质形成的拉曼光谱进行处理和识别，实现对物质成分及其比例的识别，相较于传统拉曼光谱识别技术，不需要提取待检测物质的拉曼光谱的特征峰，从而不易受到拉曼光谱中噪声的干扰，以及不会因为数据库的扩大而影响识别速度。 此外，将拉曼光谱检测设备的数据库部署在云端，拉曼光谱检测设备侧不设置数据库，在保证数据库安全性的同时降低拉曼光谱检测设备的成本。
本申请实施例基于网络，云计算，深度学习等技术，即基于云端的多任务学习的物质识别模型，对待检测物质的拉曼光谱进行物质成分识别和物质成分比例识别，其中，物质成分识别和物质成分比例识别两个识别任务在同一物质识别模型中分离进行，从而达到识别架构简单、模型复用率高以及不易受数据库扩大影响识别速度的技术效果。
为了便于本申请的实施，下面实例进行说明。
实施例1
图1示出了本申请实施例一中基于拉曼光谱的物质识别方法原理图，图2示出了本申请实施例一中基于拉曼光谱的物质识别方法流程示意图，如图1、图2所示，该方法包括：
步骤101：接收待检测物质的拉曼光谱数据。
步骤102：基于预设的多任务学习的物质识别模型，对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例。
在步骤101中，拉曼光谱采集终端(即，拉曼光谱检测设备)对待检测物质进行测量得到拉曼光谱数据，并将拉曼光谱数据通过拉曼光谱采集终端的数据传输模块经由网络发送给云端***的识别服务器，识别服务器接收待检测物质的拉曼光谱数据。
在步骤102中，识别服务器对接收到的待检测物质的拉曼光谱数据进行识别，将得到的待检测物质的物质成分及其比例的识别结果发送给拉曼光谱采集终端并显示该识别结果。
在本实施例中，所述预设的多任务学习的物质识别模型的建立，包括：
将多组单一物质的拉曼光谱数据进行多种成分比例的组合，得到组合的拉曼光谱数据；
根据组合的拉曼光谱数据的物质成分及其比例，对初始化的物质识别模型进行训练，得到训练好的多任务学习的物质识别模型。
实施中，训练服务器基于数据库中初始的拉曼光谱数据训练多任务学习的物质识别模型，并将训练好的多任务学习的物质识别模型部署到识别服务器，而不需要将数据库部署到识别服务器。
训练服务器基于数据库中初始的拉曼光谱数据训练多任务学习的物质识别模型的具体实现方法为：
1)对数据库中的所有单一物质的拉曼光谱数据进行不同成分比例的组合，形成新的拉曼光谱数据。即，对当前数据库中所有单一物质的拉曼光谱数据进行遍历，生成由多个单一物质的拉曼光谱数据组成的物质成分组合，并对所生成的包含多组拉曼光谱数据的组合分别进行全排列，组合中各物质成分比例由0开始，以0.05为步长增长至1，得到不同的物质成分比例组合。
具体为，对当前数据库中的所有单一物质的拉曼光谱数据进行两两全排列，组成所有物质的拉曼光谱数据的两两组合，例如，物质A和物质B的拉曼光谱数据的组合，以不同的成分比例进行配比，成分比例由0开始，以0.05为步长增长至1，共20种不同的成分比例组合，即物质A和物质B以0％和100％、5％和95％、10％和90％等，依次类推直至成分比例为0％和100％。
2)利用搭载深度学习框架，以及相应的图形处理器(GPU：Graphics Processing Unit)或现场可编程门阵列(FPGA：Field Programmable Gate Array)等硬件的训练服务器，将新的拉曼光谱数据作为训练样本对初始化的物质识别模型进行训练，得到训练好的多任务学习的物质识别模型，以实现对待检测物质的物质成分及其比例的识别。
在本实施例中，所述预设的多任务学习的物质识别模型中的多任务包 括用于物质成分识别的第一任务和用于物质成分比例识别的第二任务。
在本实施例中，所述预设的多任务学习的物质识别模型的损失函数的计算公式为：
损失函数＝0.5*物质成分损失函数+0.5*物质成分比例损失函数。
实施中，多任务学习的物质识别模型中的第一任务的优化目标为物质成分，第二任务的优化目标为物质成分比例。当第一任务和第二任务同时作为多任务学习的物质识别模型的优化目标时，结合第一任务的物质成分损失函数，第二任务的物质成分比例损失函数，得到新的优化目标，即将第一任务和第二任务的损失函数以1：1的比例组合成新的损失函数，并以最小化该新的损失函数作为多任务学习的物质识别模型的优化目标。
在本实施例中，所述对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例，包括：
基于第一任务识别得到多个物质成分编号以及所述物质成分对应的置信度；
根据所述物质成分对应的置信度，确定待检测物质包含的各物质成分编号；
根据所述各物质成分编号，基于第二任务识别所述各物质成分所占的比例。
实施中，将待检测物质的拉曼光谱数据作为识别服务器中多任务学习的物质识别模型的输入，通过提取拉曼光谱数据的特征向量，并基于第一任务识别得到待检测物质可能包含的各物质成分编号以及各物质成分对应的置信度，若各物质成分对应的置信度达到第一任务中设定的各物质成分的置信度阈值，例如，包含两种物质成分编号W
0和W
1，置信度分别对应为P
0和P
1，两种物质成分对应的置信度阈值为P
t0和P
t1，则确定待检测物质所含有的物质成分编号为W
0和W
1，基于第二任务对确定的待检测物质 的物质成分比例进行识别；若各物质成分对应的置信度没有全部达到第一任务中设定的各物质成分的置信度阈值，例如，包含三种物质成分编号W
0、W
1和W
2，置信度分别对应为P
0、P
1和P
2，三种物质成分对应的置信度阈值为P
t0、P
t1和P
t2，三种物质成分对应的置信度中至少一个小于对应的置信度阈值，则确定待检测物质中没有小于置信度阈值的物质成分，即假设物质成分编号W
2对应的置信度P
2小于对应的置信度阈值P
t2，则确定待检测物质所含有的物质成分编号为W
0和W
1，基于第二任务对确定的待检测物质的物质成分比例进行识别。
在本实施例中，还包括：
根据待检测物质包含的各物质成分编号，获取所述各物质成分信息。
实施中，为保证数据库中拉曼光谱数据***露，将数据库部署在训练服务器中，数据库为所有的拉曼光谱数据配置物质成分信息和对应的物质成分编号，以使数据库将物质成分编号和对应的物质成分信息发送给识别服务器，也可以在识别服务器识别得到待检测物质包含的各物质成分编号后，根据各物质成分编号访问数据库，并获取数据库中与各物质成分编号对应的物质成分信息。
本申请以具体场景为例，对本申请实施例1进行详细描述。
本申请实施例应用范围包括但不限于基于拉曼光谱的混合物质识别，以基于拉曼光谱的混合物质识别为例，基于拉曼光谱的物质识别云端***包括拉曼光谱采集终端、识别服务器和训练服务器，具体流程如下：
多任务学习的物质识别模型的训练过程：
步骤201：对数据库中的所有单一物质的拉曼光谱数据进行不同成分比例的组合，形成新的拉曼光谱数据。
步骤202：利用训练服务器，将新的拉曼光谱数据作为训练样本对初始化的物质识别模型进行训练，得到训练好的多任务学习的物质识别模型， 用于对待检测物质进行物质成分及其比例的识别。
步骤203：将训练好的多任务学习的物质识别模型部署在识别服务器中。
基于训练好的多任务学习的物质识别模型的识别过程：
步骤204：拉曼光谱采集终端对待检测物质进行测量得到拉曼光谱数据，对拉曼光谱数据进行定标、去基底噪声等处理后上传至云端***的识别服务器。
步骤205：识别服务器对接收到的待检测物质的拉曼光谱数据进行归一化等处理，归一化处理具体为，对拉曼光谱数据的x轴波数的范围和分辨率进行归一化处理。
步骤206：将归一化处理后的待检测物质的拉曼光谱数据输入多任务学习的物质识别模型，通过提取拉曼光谱数据的特征向量，并基于第一任务输出待检测物质可能包含的各物质成分编号以及各物质成分对应的置信度，根据各物质成分对应的置信度阈值确定待检测物质包含的各物质成分编号，以及基于第二任务输出待检测物质中各物质成分所占的比例。
步骤207：根据待检测物质中各物质成分编号访问数据库，并获取数据库中与各物质成分编号对应的物质成分信息。
步骤208：将待检测物质的物质成分信息及物质成分比例发送给拉曼光谱采集终端，以使拉曼光谱采集终端显示该检测物质的物质成分信息及物质成分比例。
实施例2
基于同一发明构思，本申请实施例中还提供了一种基于拉曼光谱的物质识别云端***，由于这些设备解决问题的原理与一种基于拉曼光谱的物质识别方法相似，因此这些设备的实施可以参见方法的实施，重复之处不再赘述。
图3示出了本申请实施例二中基于拉曼光谱的物质识别云端***结构图，如图3所示，基于拉曼光谱的物质识别云端***300可以包括：
拉曼光谱采集终端301，用于采集待检测物质的拉曼光谱数据。
识别服务器302，用于接收待检测物质的拉曼光谱数据；以及，
基于预设的多任务学习的物质识别模型，对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例。
训练服务器303，用于所述预设的多任务学习的物质识别模型的建立，所述训练服务器303包括：
将多组单一物质的拉曼光谱数据进行多种成分比例的组合，得到组合的拉曼光谱数据；
根据组合的拉曼光谱数据的物质成分及其比例，对初始化的物质识别模型进行训练，得到训练好的多任务学习的物质识别模型。
在本实施例中，所述预设的多任务学习的物质识别模型中的多任务包括用于物质成分识别的第一任务和用于物质成分比例识别的第二任务。
在本实施例中，所述预设的多任务学习的物质识别模型的损失函数的计算公式为：
损失函数＝0.5*物质成分损失函数+0.5*物质成分比例损失函数。
在本实施例中，所述对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例，包括：
基于第一任务识别得到多个物质成分编号以及所述物质成分对应的置信度；
根据所述物质成分对应的置信度，确定待检测物质包含的各物质成分编号；
根据所述各物质成分编号，基于第二任务识别所述各物质成分所占的比例。
在本实施例中，还包括：
根据待检测物质包含的各物质成分编号，获取所述各物质成分信息。
实施例3
基于同一发明构思，本申请实施例中还提供了一种电子设备，由于其原理与一种基于拉曼光谱的物质识别方法相似，因此其实施可以参见方法的实施，重复之处不再赘述。
图4示出了本申请实施例三中电子设备的结构示意图，如图4所示，所述电子设备包括：收发设备401，存储器402，一个或多个处理器403；以及一个或多个模块，所述一个或多个模块被存储在所述存储器中，并被配置成由所述一个或多个处理器执行，所述一个或多个模块包括用于执行任一上述方法中各个步骤的指令。
实施例4
基于同一发明构思，本申请实施例还提供了一种与电子设备结合使用的计算机程序产品，由于其原理与一种基于拉曼光谱的物质识别方法相似，因此其实施可以参见方法的实施，重复之处不再赘述。所述计算机程序产品包括计算机可读的存储介质和内嵌于其中的计算机程序机制，所述计算机程序机制包括用于执行任一上述方法中各个步骤的指令。
为了描述的方便，以上所述装置的各部分以功能分为各种模块分别描述。当然，在实施本申请时可以把各模块或单元的功能在同一个或多个软件或硬件中实现。
本领域内的技术人员应明白，本申请的实施例可提供为方法、***、或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的 形式。
本申请是参照根据本申请实施例的方法、设备(***)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
尽管已描述了本申请的优选实施例，但本领域内的技术人员一旦得知了基本创造性概念，则可对这些实施例作出另外的变更和修改。所以，所附权利要求意欲解释为包括优选实施例以及落入本申请范围的所有变更和修改。
Claims (14)
- 一种基于拉曼光谱的物质识别方法，其特征在于，包括：接收待检测物质的拉曼光谱数据；基于预设的多任务学习的物质识别模型，对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例。
- 如权利要求1所述的方法，其特征在于，所述预设的多任务学习的物质识别模型的建立，包括：将多组单一物质的拉曼光谱数据进行多种成分比例的组合，得到组合的拉曼光谱数据；根据组合的拉曼光谱数据的物质成分及其比例，对初始化的物质识别模型进行训练，得到训练好的多任务学习的物质识别模型。
- 如权利要求1所述的方法，其特征在于，所述预设的多任务学习的物质识别模型中的多任务包括用于物质成分识别的第一任务和用于物质成分比例识别的第二任务。
- 如权利要求1或3所述的方法，其特征在于，所述预设的多任务学习的物质识别模型的损失函数的计算公式为：损失函数＝0.5*物质成分损失函数+0.5*物质成分比例损失函数。
- 如权利要求3所述的方法，其特征在于，所述对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例，包括：基于第一任务识别得到多个物质成分编号以及所述物质成分对应的置信度；根据所述物质成分对应的置信度，确定待检测物质包含的各物质成分编号；根据所述各物质成分编号，基于第二任务识别所述各物质成分所占的比例。
- 如权利要求5所述的方法，其特征在于，还包括：根据待检测物质包含的各物质成分编号，获取所述各物质成分信息。
- 一种基于拉曼光谱的物质识别云端***，其特征在于，包括：拉曼光谱采集终端，用于采集待检测物质的拉曼光谱数据；识别服务器，用于接收待检测物质的拉曼光谱数据；以及，基于预设的多任务学习的物质识别模型，对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例。
- 如权利要求7所述的云端***，其特征在于，还包括：训练服务器，用于所述预设的多任务学习的物质识别模型的建立，所述训练服务器包括：将多组单一物质的拉曼光谱数据进行多种成分比例的组合，得到组合的拉曼光谱数据；根据组合的拉曼光谱数据的物质成分及其比例，对初始化的物质识别模型进行训练，得到训练好的多任务学习的物质识别模型。
- 如权利要求7所述的云端***，其特征在于，所述预设的多任务学习的物质识别模型中的多任务包括用于物质成分识别的第一任务和用于物质成分比例识别的第二任务。
- 如权利要求7或9所述的云端***，其特征在于，所述预设的多任务学习的物质识别模型的损失函数的计算公式为：损失函数＝0.5*物质成分损失函数+0.5*物质成分比例损失函数。
- 如权利要求9所述的云端***，其特征在于，所述对所述待检测物质的拉曼光谱进行识别，得到所述待检测物质的物质成分及其比例，包括：基于第一任务识别得到多个物质成分编号以及所述物质成分对应的置信度；根据所述物质成分对应的置信度，确定待检测物质包含的各物质成分 编号；根据所述各物质成分编号，基于第二任务识别所述各物质成分所占的比例。
- 如权利要求11所述的云端***，其特征在于，还包括：根据待检测物质包含的各物质成分编号，获取所述各物质成分信息。
- 一种电子设备，其特征在于，所述电子设备包括：收发设备，存储器，一个或多个处理器；以及一个或多个模块，所述一个或多个模块被存储在所述存储器中，并被配置成由所述一个或多个处理器执行，所述一个或多个模块包括用于执行权利要求1-6中任一所述方法中各个步骤的指令。
- 一种与电子设备结合使用的计算机程序产品，所述计算机程序产品包括计算机可读的存储介质和内嵌于其中的计算机程序机制，所述计算机程序机制包括用于执行权利要求1-6中任一所述方法中各个步骤的指令。
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201780002761.2A CN108235733B (zh) | 2017-12-29 | 2017-12-29 | 基于拉曼光谱的物质识别方法及云端*** |
| PCT/CN2017/119796 WO2019127352A1 (zh) | 2017-12-29 | 2017-12-29 | 基于拉曼光谱的物质识别方法及云端*** |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2017/119796 WO2019127352A1 (zh) | 2017-12-29 | 2017-12-29 | 基于拉曼光谱的物质识别方法及云端*** |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2019127352A1 true WO2019127352A1 (zh) | 2019-07-04 |
Family
ID=62643329
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2017/119796 WO2019127352A1 (zh) | 2017-12-29 | 2017-12-29 | 基于拉曼光谱的物质识别方法及云端*** |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN108235733B (zh) |
| WO (1) | WO2019127352A1 (zh) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113008865A (zh) * | 2021-03-04 | 2021-06-22 | 北京华泰诺安探测技术有限公司 | 珠宝玉石鉴定方法、装置、介质及设备 |
| CN117907309A (zh) * | 2024-03-19 | 2024-04-19 | 夏芮智能科技有限公司 | 一种基于拉曼光谱的食品和药品安全检测*** |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020024167A1 (zh) * | 2018-08-01 | 2020-02-06 | 深圳达闼科技控股有限公司 | 一种物质检测数据库的数据添加方法、装置和检测设备 |
| CN109001180B (zh) * | 2018-08-10 | 2021-01-01 | 青岛启明生物科技有限公司 | 一种拉曼光谱结合人工智能高通量单细胞分析鉴定方法 |
| CN113466206A (zh) * | 2021-06-23 | 2021-10-01 | 上海仪电（集团）有限公司中央研究院 | 一种基于大数据的拉曼光谱分析*** |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140088876A1 (en) * | 2012-09-26 | 2014-03-27 | Panalytical Inc | Multi-sensor analysis of complex geologic materials |
| CN104215623A (zh) * | 2013-05-31 | 2014-12-17 | 欧普图斯(苏州)光学纳米科技有限公司 | 面向多行业检测的激光拉曼光谱智能化辨识方法及*** |
| CN105954258A (zh) * | 2016-05-26 | 2016-09-21 | 伯格森(北京)科技有限公司 | 食用油掺劣质油检测仪及检测方法 |
| CN106323889A (zh) * | 2016-08-11 | 2017-01-11 | 上海赛诚医药科技有限公司 | 基于云服务器的质量检测*** |
| CN106706546A (zh) * | 2016-12-28 | 2017-05-24 | 中山市腾创贸易有限公司 | 一种基于红外和拉曼光谱数据的人工智能学习物质分析方法 |
| CN107037028A (zh) * | 2017-03-10 | 2017-08-11 | 北京华泰诺安探测技术有限公司 | 一种云平台拉曼光谱识别方法及装置 |
Family Cites Families (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE10051806A1 (de) * | 2000-10-18 | 2002-04-25 | Bayer Ag | Verfahren zur Charakterisierung, Identifizierung und Kennzeichnung von mikrobiellen Mischungen |
| US7410763B2 (en) * | 2005-09-01 | 2008-08-12 | Intel Corporation | Multiplex data collection and analysis in bioanalyte detection |
| US8279437B2 (en) * | 2010-07-14 | 2012-10-02 | Hewlett-Packard Development Company, L.P. | Multi-wavelength Raman light detection for detecting a species |
| EP2525213A1 (en) * | 2011-05-16 | 2012-11-21 | Renishaw plc | Spectroscopic apparatus and methods for determining components present in a sample |
| CN102306236B (zh) * | 2011-07-06 | 2015-04-29 | 广西工学院 | 基于角度度量的多变量分析方法 |
| CN103217411B (zh) * | 2013-04-15 | 2015-03-25 | 浙江大学 | 基于拉曼光谱的食用植物油中油酸、亚油酸和饱和脂肪酸含量检测方法 |
| CN103398970B (zh) * | 2013-07-24 | 2015-09-16 | 骆驰 | 一种定性和定量分析食用油并进而检测地沟油的方法 |
| CN104458703B (zh) * | 2014-12-16 | 2017-07-21 | 盐城工学院 | 一种转基因水稻种子及其亲本的快速检测方法及其专用装置 |
| CN105651705A (zh) * | 2015-01-14 | 2016-06-08 | 青海春天药用资源科技利用有限公司 | 无损检测冬虫夏草粉/粉片中伪品含量的方法 |
| CN104792752A (zh) * | 2015-04-03 | 2015-07-22 | 江南大学 | 三维荧光光谱结合parafac算法测定混合色素溶液中色素含量的方法 |
| CN104764734A (zh) * | 2015-04-30 | 2015-07-08 | 江苏易谱恒科技有限公司 | 基于拉曼光谱和svm算法的果汁香精香料的鉴别方法 |
| CN105223184A (zh) * | 2015-10-23 | 2016-01-06 | 上海卫华科学仪器有限公司 | 基于拉曼光谱仪的物质定性及其组分定量检测方法 |
| CN105588828A (zh) * | 2015-12-16 | 2016-05-18 | 新希望双喜乳业（苏州）有限公司 | 一种拉曼光谱快速检测牛奶脂肪含量的方法 |
| CN105548079A (zh) * | 2015-12-28 | 2016-05-04 | 浙江中烟工业有限责任公司 | 一种基于近红外光谱的烟丝组成测定方法 |
| CN105486675B (zh) * | 2015-12-29 | 2018-07-03 | 长安大学 | 定量分析co2气体碳同位素组成的激光拉曼检测方法 |
| CN105954252B (zh) * | 2016-04-21 | 2018-10-02 | 北京航空航天大学 | 一种饲料原料中非法添加物苏丹红的定性检测方法 |
| CN105806825B (zh) * | 2016-05-17 | 2019-07-23 | 浙江大学 | 一种天然气组分在线气体拉曼分析方法 |
| CN106124449B (zh) * | 2016-06-07 | 2019-03-05 | 中国科学院合肥物质科学研究院 | 一种基于深度学习技术的土壤近红外光谱分析预测方法 |
| CN106127204B (zh) * | 2016-06-30 | 2019-08-09 | 华南理工大学 | 一种全卷积神经网络的多方向水表读数区域检测算法 |
| CN106951867B (zh) * | 2017-03-22 | 2019-08-23 | 成都擎天树科技有限公司 | 基于卷积神经网络的人脸识别方法、装置、***及设备 |
| CN107451602A (zh) * | 2017-07-06 | 2017-12-08 | 浙江工业大学 | 一种基于深度学习的果蔬检测方法 |
-
2017
- 2017-12-29 WO PCT/CN2017/119796 patent/WO2019127352A1/zh active Application Filing
- 2017-12-29 CN CN201780002761.2A patent/CN108235733B/zh active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140088876A1 (en) * | 2012-09-26 | 2014-03-27 | Panalytical Inc | Multi-sensor analysis of complex geologic materials |
| CN104215623A (zh) * | 2013-05-31 | 2014-12-17 | 欧普图斯(苏州)光学纳米科技有限公司 | 面向多行业检测的激光拉曼光谱智能化辨识方法及*** |
| CN105954258A (zh) * | 2016-05-26 | 2016-09-21 | 伯格森(北京)科技有限公司 | 食用油掺劣质油检测仪及检测方法 |
| CN106323889A (zh) * | 2016-08-11 | 2017-01-11 | 上海赛诚医药科技有限公司 | 基于云服务器的质量检测*** |
| CN106706546A (zh) * | 2016-12-28 | 2017-05-24 | 中山市腾创贸易有限公司 | 一种基于红外和拉曼光谱数据的人工智能学习物质分析方法 |
| CN107037028A (zh) * | 2017-03-10 | 2017-08-11 | 北京华泰诺安探测技术有限公司 | 一种云平台拉曼光谱识别方法及装置 |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113008865A (zh) * | 2021-03-04 | 2021-06-22 | 北京华泰诺安探测技术有限公司 | 珠宝玉石鉴定方法、装置、介质及设备 |
| CN117907309A (zh) * | 2024-03-19 | 2024-04-19 | 夏芮智能科技有限公司 | 一种基于拉曼光谱的食品和药品安全检测*** |
| CN117907309B (zh) * | 2024-03-19 | 2024-06-04 | 夏芮智能科技有限公司 | 一种基于拉曼光谱的食品和药品安全检测*** |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108235733A (zh) | 2018-06-29 |
| CN108235733B (zh) | 2022-01-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2019127352A1 (zh) | 基于拉曼光谱的物质识别方法及云端*** | |
| Truong et al. | Machine learning cryptanalysis of a quantum random number generator | |
| US11493447B2 (en) | Method for removing background from spectrogram, method of identifying substances through Raman spectrogram, and electronic apparatus | |
| US10267678B2 (en) | Raman spectroscopic detection method | |
| US8826092B2 (en) | Characterization and validation of processor links | |
| JP2018185328A5 (zh) | ||
| Thrane et al. | Detecting Gravitational-Wave Transients at 5 σ: A Hierarchical Approach | |
| CN108780048A (zh) | 一种确定检测设备的方法、检测装置及可读存储介质 | |
| US10408760B2 (en) | Raman spectrum detecting method for eliminating package interference and electronic device thereof | |
| Soni et al. | Malicious application detection in android using machine learning | |
| US11126897B2 (en) | Unification of classifier models across device platforms | |
| CN108195817B (zh) | 用于去除溶剂干扰的拉曼光谱检测方法 | |
| CN109800797A (zh) | 基于ai的文件黑白判断方法、装置及设备 | |
| Robyns et al. | Practical operation extraction from electromagnetic leakage for side-channel analysis and reverse engineering | |
| CN106936561B (zh) | 一种侧信道攻击防护能力评估方法和*** | |
| US10852229B2 (en) | Detection of compounds in material samples utilizing a transformation and subsequent decomposition | |
| CN111157115B (zh) | 一种水下布里渊散射光谱获取方法及装置 | |
| EP3561718A1 (en) | Method for identifying raman spectrogram, and electronic device | |
| Sulis et al. | Using hydrodynamical simulations of stellar atmospheres for periodogram standardization: Application to exoplanet detection | |
| Mu et al. | Fluorescence polarization technique: a new method for vegetable oils classification | |
| US20170109598A1 (en) | Method and apparatus for user authentication using raman spectrum | |
| US9069849B1 (en) | Methods for enforcing time alignment for speed resistant audio matching | |
| Trautmann et al. | Characterization of Side Channels on FPGA-based Off-The-Shelf Boards against Automated Attacks | |
| WO2019227291A1 (zh) | 基于拉曼光谱和libs的物质检测方法、装置及电子设备 | |
| FR3048086A1 (fr) | Procede de test de la resistance d'un circuit a une analyse par canaux auxiliaires |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 17936336 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 16/11/2020) |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 17936336 Country of ref document: EP Kind code of ref document: A1 |