CN111609931A - Parameter-programmable real-time hyperspectral acquisition system and method - Google Patents

Abstract

The invention discloses a parameter programmable real-time hyperspectral collection method, which comprises the following steps: the method comprises the steps that each frame of data stored in an SSD is moved to a DDR array by a program, the data are preprocessed and cut, the data are put into a multi-core GPU to be processed in parallel by using a preset intelligent algorithm, processed real-time reports are stored in the SSD connected to a GPU end, finally a user checks the data in a web browser or the APP, each frame of hyperspectral data are preprocessed by an onboard GPU in a necessary parallel mode, then the hyperspectral data are placed in a cache array, and the data are transmitted to a GPU array group at a cloud server end to be processed through a 5G network. According to the hyperspectral image data acquisition, storage and transmission terminal taking the GPU and the FPGA as the core, the target spectrum channel can be selected at will, and the multispectral image data acquisition, storage and transmission terminal can be switched between multispectral and hyperspectral modes according to the measurement intention and occasions, so that data redundancy is greatly reduced, the data processing workload is reduced, the data storage pressure is reduced, and the operation efficiency is improved.

Description

Parameter-programmable real-time hyperspectral acquisition system and method

Technical Field

The invention relates to the technical field of remote sensing measurement, in particular to a parameter programmable real-time hyperspectral acquisition system and method.

Background

In the hyperspectral remote sensing measurement, a plurality of mutually opposite requirements exist all the time, and the requirements are difficult to be considered or cannot be realized at the same time. For the high spectrum band number, when data is acquired, the target spectrum information is expected to be acquired fully and completely, so that the data volume is large, and high spectrum data has great redundant information; in the study of an unknown object, it is desirable to use a sufficient number of bands and a sufficiently comprehensive spectrum of information; in actual operation, because the spectrum information of the observation target is known, only specific wave bands are needed, namely, one customized multichannel spectral device can complete the operation, the cost is low, the data volume is small, the data redundancy is small, and the analysis and the processing of the data are facilitated. Most of common hyperspectral equipment on the market is fixed in wave band, and less wave bands can be flexibly configured; therefore, if the equipment can conveniently and freely configure the wave band, the data storage pressure is reduced, the data redundancy is reduced, and the data processing working efficiency is obviously improved.

Present high spectrum measuring equipment all does not possess on-the-spot real-time or quasi-real-time processing function, all needs the data processing through the later stage just can use, very big restriction high spectral application scene, can not use in the above-mentioned occasion that requires highly to the real-time nature, along with the rise of 5G technique, higher-speed transmission bandwidth for data long-range back cloud server computer lab real-time processing becomes possible, this storage and the computational pressure that will alleviate terminal equipment.

Disclosure of Invention

The invention aims to provide a parameter programmable real-time hyperspectral acquisition system and method to solve the problems in the background technology.

In order to achieve the purpose, the invention provides the following technical scheme: the utility model provides a parameter real-time hyperspectral collection system that is programmable, includes GPU, GPU passes through high-speed PCIE interface and RS232 serial ports and FPGA signal connection, GPU's input and USB spectrum camera's output end signal connection, FPGA's input and the output end signal connection of camera alink interface spectrum camera, the input of camera alink interface spectrum camera and algorithm correction module's output end signal connection, GPU's output and the input signal connection of WIFI bluetooth 4G 5G module, the output of WIFI bluetooth 4G 5G module and the input signal connection of user side.

Preferably, the user side includes a web browser and an APP inside.

A parameter programmable real-time hyperspectral collection method comprises the following steps:

s1, moving each frame of data stored in the SSD to a DDR array by the program, preprocessing and cutting the data, putting the data into a multi-core GPU, performing parallel processing by using a preset intelligent algorithm, and collecting and processing the data at the same time;

s2, storing the processed real-time report in an SSD connected to a GPU end, and finally checking the report in a web browser or an APP by a user through a WIFI/Bluetooth/4G/5G module;

a parameter programmable real-time hyperspectral collection method comprises the following steps:

s1, performing necessary parallel data preprocessing on each frame of hyperspectral data through an onboard GPU, placing the hyperspectral data in a cache array, and transmitting the hyperspectral data to a GPU array group at a cloud server end through a 5G network for processing;

and S2, collecting and processing the data, transmitting the processed result to the SSD connected with the GPU through the 5G network, and finally checking the result in the web browser or APP of the mobile equipment by the user.

Compared with the prior art, the invention has the beneficial effects that:

(1) according to the terminal for hyperspectral image data acquisition, storage and transmission, which takes the GPU and the FPGA as the core, a target spectrum channel can be selected at will, and the terminal can be switched between multispectral and hyperspectral modes according to measurement intents and occasions, so that data redundancy is greatly reduced, data processing workload is reduced, data storage pressure is reduced, operation efficiency is improved, and the GPU supports interpretation algorithm customization and algorithm model importing functions;

(2) according to the invention, each terminal is configured with a group of onboard high-speed multi-core GPU modules, so that real-time hyperspectral image interpretation of the terminal can be realized, a single terminal can be invited to realize networking of a plurality of terminals into a cloud processor, and meanwhile, the single terminal can be connected with a GPU array at the cloud processor end of a remote computer room through a 5G network to realize real-time hyperspectral processing;

(3) according to the method, each frame of data stored in the SSD is moved to the DDR array, the data is preprocessed and cut, the data is put into the multi-core GPU for parallel processing by using a preset intelligent algorithm, the data is collected and processed at the same time, a processed real-time report is stored in the SSD connected to the GPU end, and a user can check the report in real time in a web browser or an APP of the mobile device.

Drawings

FIG. 1 is a system diagram of the present invention;

FIG. 2 is a block diagram of a GPU real-time processing of the present invention;

FIG. 3 is a block diagram of a 5G-based hyperspectral cloud real-time processing system.

In the figure: the system comprises a GPU (graphics processing unit), a FPGA (field programmable gate array), a 3USB (universal serial bus) spectral camera, a 4Cameralink interface spectral camera, a 5 algorithm correction module, a 6 WIFI/Bluetooth/4G/5G module, a 7 user side, a 71web browser and a 72 APP.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The first embodiment is as follows:

the utility model provides a parameter real-time hyperspectral collection system that is programmable, includes GPU1, GPU1 is through high-speed PCIE interface and RS232 serial ports and FPGA2 signal connection, GPU 1's input and USB spectrum camera 3's output end signal connection, FPGA 2's input and Cameralink interface spectrum camera 4's output end signal connection, Cameralink interface spectrum camera 4's input and algorithm correction module 5's output end signal connection, GPU 1's output and WIFI/bluetooth/4G 5G module 6's input end signal connection, WIFI/bluetooth/4G 5G module 6's output and user terminal 7's input end signal connection, user terminal 7's inside includes web browser 71 and APP 72.

A parameter programmable real-time hyperspectral collection method comprises the following steps:

s1, moving each frame of data stored in the SSD to a DDR array by the program, preprocessing and cutting the data, putting the data into a multi-core GPU, performing parallel processing by using a preset intelligent algorithm, and collecting and processing the data at the same time;

s2, storing the processed real-time report in an SSD connected to a GPU end, and finally checking the report in a web browser or APP by a user through a WIFI/Bluetooth/4G/5G module.

Example two:

the utility model provides a parameter real-time hyperspectral collection system that is programmable, includes GPU1, GPU1 is through high-speed PCIE interface and RS232 serial ports and FPGA2 signal connection, GPU 1's input and USB spectrum camera 3's output end signal connection, FPGA 2's input and Cameralink interface spectrum camera 4's output end signal connection, Cameralink interface spectrum camera 4's input and algorithm correction module 5's output end signal connection, GPU 1's output and WIFI/bluetooth/4G 5G module 6's input end signal connection, WIFI/bluetooth/4G 5G module 6's output and user terminal 7's input end signal connection, user terminal 7's inside includes web browser 71 and APP 72.

A parameter programmable real-time hyperspectral collection method comprises the following steps:

s1, performing necessary parallel data preprocessing on each frame of hyperspectral data through an onboard GPU, placing the hyperspectral data in a cache array, and transmitting the hyperspectral data to a GPU array group at a cloud server end through a 5G network for processing;

and S2, collecting and processing the data, transmitting the processed result to the SSD connected with the GPU through the 5G network, and finally checking the result in the web browser or APP of the mobile equipment by the user.

The working principle is as follows: when the method is used, each frame of data stored in the SSD is moved to a DDR array by a program, the data is preprocessed and cut, the data is put into a multi-core GPU to be processed in parallel by using a preset intelligent algorithm, the data is collected and processed at the same time, a processed real-time report is stored in an SSD connected to a GPU end and then passes through a WIFI/Bluetooth/4G/5G module, finally a user checks the data in a web browser or the APP, each frame of hyperspectral data is preprocessed by necessary parallel data through an on-board GPU, then the hyperspectral data is put into a cache array, the hyperspectral data is transmitted to a GPU array group at a cloud server end through a 5G network to be processed at the same time, a processed result is transmitted to the SSD connected with the GPU through the 5G network, and finally the user checks the processed result in the web browser or the APP of mobile.

In summary, the terminal for collecting, storing and transmitting the hyperspectral image data by taking the GPU1 and the FPGA2 as the core can switch between multispectral and hyperspectral modes according to measurement intents and occasions, greatly reduces data redundancy, reduces data processing workload, relieves data storage pressure and improves operation efficiency, real-time hyperspectral image interpretation of the terminal can be realized by configuring a group of onboard high-speed multi-core GPU modules for each terminal, a plurality of terminals can be networked into a cloud processor by a single terminal in an inviting manner, meanwhile, real-time hyperspectral processing can be realized by connecting a GPU array at the cloud processor end of a remote computer room through a 5G network, each frame of data stored in an SSD is moved to the DDR array, and is put into a multi-core GPU for parallel processing by using a preset intelligent algorithm after data preprocessing and cutting, while collecting and processing, the processed real-time report is stored in an SSD connected to the GPU1 end, and a user can view the report in real time in the web browser 71 or APP72 of the mobile device.

Claims (4)

1. A parameter-programmable real-time hyperspectral acquisition system comprises a GPU (1), and is characterized in that: the system comprises a GPU (1), a USB spectrum camera (3), a Cameralink interface spectrum camera (4), a high-speed PCIE interface, an RS232 serial port, a Field Programmable Gate Array (FPGA) (2), a high-speed PCIE interface, an RS232 serial port, an input end of the GPU (1), an input end of the FPGA (2), an output end of the Cameralink interface spectrum camera (4), an input end of the Cameralink interface spectrum camera (4), an output end of the GPU (1), an input end of a WIFI/Bluetooth/4G/5G module (6), and an output end of the WIFI/Bluetooth/4G/5G module (6), wherein the output end of the WIFI/Bluetooth/4G/5G module (6) is in signal connection.

2. The parameter programmable real-time hyperspectral acquisition system of claim 1, wherein: the user terminal (7) comprises a web browser (71) and an APP (72) inside.

3. A parameter programmable real-time hyperspectral collection method is characterized in that a first method comprises the following steps:

s1, moving each frame of data stored in the SSD to a DDR array by the program, preprocessing and cutting the data, putting the data into a multi-core GPU, performing parallel processing by using a preset intelligent algorithm, and collecting and processing the data at the same time;

s2, storing the processed real-time report in an SSD connected to a GPU end, and finally checking the report in a web browser or APP by a user through a WIFI/Bluetooth/4G/5G module.

4. A parameter programmable real-time hyperspectral collection method is characterized in that a second method comprises the following steps:

s1, performing necessary parallel data preprocessing on each frame of hyperspectral data through an onboard GPU, placing the hyperspectral data in a cache array, and transmitting the hyperspectral data to a GPU array group at a cloud server end through a 5G network for processing;

and S2, collecting and processing the data, transmitting the processed result to the SSD connected with the GPU through the 5G network, and finally checking the result in the web browser or APP of the mobile equipment by the user.

Images