CN107292809A - A kind of method that GPU realizes ultrasonic signal filtering process - Google Patents
A kind of method that GPU realizes ultrasonic signal filtering process Download PDFInfo
- Publication number
- CN107292809A CN107292809A CN201610586805.8A CN201610586805A CN107292809A CN 107292809 A CN107292809 A CN 107292809A CN 201610586805 A CN201610586805 A CN 201610586805A CN 107292809 A CN107292809 A CN 107292809A
- Authority
- CN
- China
- Prior art keywords
- signal
- ultrasonic signal
- gpu
- cache space
- filtering process
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/20—Processor architectures; Processor configuration, e.g. pipelining
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/5207—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving processing of raw data to produce diagnostic data, e.g. for generating an image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/94—Hardware or software architectures specially adapted for image or video understanding
- G06V10/955—Hardware or software architectures specially adapted for image or video understanding using specific electronic processors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
Abstract
Technical scheme includes a kind of method that GPU realizes ultrasonic signal filtering process, and the system includes:S1, opens up source cache space, target cache space and temporal cache space, loads the operational order of signal transacting;S2, obtains ultrasonic signal, and calculating obtains signal ephemeral data;S3, obtaining step S2 ephemeral data obtains signal results data;S4, circulation performs step S2, S3, and all obtained echo signals are carried out imaging by all signal ways of traversal loop.Beneficial effects of the present invention are:Idle GPU resource can effectively be utilized, it is to avoid computer resource is wasted;The operating structure cleverly utilized, solves the ultrasonic signal problem that prior art can not be handled using GPU;Efficient process ultrasonic signal speed.
Description
Technical field
The present invention relates to a kind of method that GPU realizes ultrasonic signal filtering process, belong to medical image imaging field.
Background technology
Ultrasonic signal is after amplification and AD collections, and follow-up filtering process is for the extremely important of quality that is finally imaged
High-order band logical dynamic filter is generally used, the requirement to operand is very big, handled so being normally placed on FPGA, to mitigate CPU
Burden however this can increase equipment volume and cost, and the flexibility of algorithm is restricted
Signal filtering processing is placed on main frame and handled, equipment volume and cost can be not only reduced, total system is improved reliable
Property, and the flexibility of algorithm can be improved.In the different working modes and working frequency of ultrasonic device, the requirement to wave filter
It is different, being realized on FPGA will realize that flexible parameter and algorithm adjustment are more difficult.But filtering is run on main frame
Algorithm, the requirement to CPU is again relatively higher.
At present, main frame all carries GPU, and GPU generally only shows for doing image.But modern GPU has had powerful
Programmability, whether through DirectX, penGL, or OpenCL, can realize GPU program.GPU computing energy
Power is generally several times in CPU, and under normal conditions, this operational capability is idle.
GPU is that have many kernels run parallel, and concurrent operation is done to mass data.On surface, filtering algorithm is all
Serial process algorithm, GPU is not appropriate for doing such task.
The content of the invention
Flexible, complicated computing can not be carried out on FPGA for prior art, technical scheme provides one
Plant the efficient processing method in GPU processing cores.
Technical scheme includes a kind of method that GPU realizes ultrasonic signal filtering process, it is characterised in that the party
Method includes:S1, opens up source cache space, target cache space and temporal cache space, while at GPU in GPU processing cores
Manage the operational order that core loads signal transacting;S2, GPU processing core obtain the i-th road ultrasonic signal, and according to operational order
Progress, which is calculated, obtains signal ephemeral data, and signal ephemeral data is sent to temporal cache space;S3, obtains and is based on step S2
The ephemeral data on the i-th obtained tunnel, computing is carried out with reference to the source ultrasonic signal on the i+1 road of acquisition, and according to operational order
And signal results data are obtained, and then signal results data are sent to target cache space;S4, circulation performs step S2, S3,
The signal results data that circulation obtains in step S3 every time are the echo signal on the i-th tunnel, until all signal ways of traversal loop
Afterwards, all obtained echo signals are subjected to imaging.
Further, the step S1 also includes:It is the consistent multichannel ultrasound of length that ultrasonic signal is carried out into preliminary exposition
Ripple signal, i value is less than N-1 to the total way N of ultrasonic signal therein between 64 and 256, and wherein.
Further, the step S1 also includes:Corresponding operational order is created according to different ultrasonic signal sources, and
It can be the combination of nonidentity operation instruction.
Further, the step S2 and S3 also includes:The echo signal form of output therein is used and input ultrasonic wave
Signal format is consistent.
Further, the step S2 and S3 also includes:GPU processing cores, will be from original caching when obtaining ultrasonic signal
The echo signal that the ultrasonic signal that space is obtained is obtained by circular treatment is distributed to the specified coordinate in target cache space, and
Wherein acquired ultrasonic signal coordinate can be set with self-defined.
Beneficial effects of the present invention are:Idle GPU resource can effectively be utilized, it is to avoid computer resource is wasted;Cleverly
The operating structure utilized, solves the ultrasonic signal problem that prior art can not be handled using GPU;Efficient process ultrasound
Ripple signal speed.
Brief description of the drawings
Fig. 1 is shown according to the usual arithmographs of embodiment of the present invention GPU;
Fig. 2 show the signal filtering processing figure according to embodiment of the present invention;
Constituted and schemed according to the ultrasonic signal data structure of embodiment of the present invention shown in Fig. 3;
Fig. 4 a, 4b show the circular treatment figure according to embodiment of the present invention.
Embodiment
In order that the object, technical solutions and advantages of the present invention are clearer, below in conjunction with the accompanying drawings with specific embodiment pair
The present invention is described in detail.The GPU of the present invention realizes that the method for ultrasonic signal filtering process is applied at medical ultrasonic signal
Reason.
Fig. 1 is shown according to the usual arithmographs of embodiment of the present invention GPU.Specify 1 or several source data buffer, 1
Individual or several purpose data buffer, operational order is loaded into after GPU, and GPU kernel can be according to allocated coordinate, from source
Data buffer reads data, result is output on purpose buffer corresponding with coordinate position after computing.
That is, output data position be with coordinate binding, and to be code can choose for the position of input data.
Fig. 2 show the signal filtering processing figure according to embodiment of the present invention.Data are sequential processes, and below
Data processing need to use result of calculation above.I.e. circulation performs step, before circulation acquisition signal results data are every time
Echo signal all the way, until after all signal ways of traversal loop, all obtained echo signals are carried out into imaging.
Constituted and schemed according to the ultrasonic signal data structure of embodiment of the present invention shown in Fig. 3.Usual ultrasonic signal is 64
Road to 256 tunnels.Needs so multichannel data is ultimately imaged all to have handled.So we are organized into signal the input of Fig. 3 forms
Buffer, while exporting buffer also uses identical form, while preparing the interim buffer of two same formats, still
Interim buffer each unit includes multiple data.
Fig. 4 a, 4b show the circular treatment figure according to embodiment of the present invention.Fig. 4 a, Fig. 4 two steps of b are pressed to GPU
It is alternately performed multiple computing, each calculation process such as Fig. 4 a, shown in Fig. 4 b.Write-in Temp Buffer packet is containing filter every time
The ephemeral data of ripple algorithm and current result data(It is identical with the data for being output to Dest Buffer).In this computing frame
Under structure, the signal parallel on all roads can be made to run any combination of filtering algorithm on GPU, including RC/LC/LR. so may be used
As long as high-pass/low-pass/band logical Butterworth/Chebyshev to realize arbitrary order etc. wave filter .GPU programs are realized corresponding
Algorithm filter code.
By taking simple 1 rank high pass LC wave filters as an example, false code is as follows:
// computing
Temp1 = (srcInput-(TempInput.result – TempInput.temp1) + k*
Tempinput.temp2)/c;
TempResult = k*(Temp1 – TempInput.temp2);
// output
Dst = TempResult;
TempOutput.temp1 = TempResult;
TempOutput.temp2 = Temp1;
In the integrated graphics core (GPU) of Intel Celeron, such tunnel of code process 256,40MHz sample frequencys, depth
20cm signal, can obtain the processing speed more than 200 FPS., can be easily more than 10000FPS on high-performance GPU.
It is described above, simply presently preferred embodiments of the present invention, the invention is not limited in above-mentioned embodiment, as long as
It reaches the technique effect of the present invention with identical means, should all belong to protection scope of the present invention.In the protection model of the present invention
Its technical scheme and/or embodiment can have a variety of modifications and variations in enclosing.
Claims (5)
1. a kind of method that GPU realizes ultrasonic signal filtering process, it is characterised in that this method includes:
S1, source cache space, target cache space and temporal cache space are opened up in GPU processing cores, while in GPU process cores
The heart loads the operational order of signal transacting;
S2, GPU processing core obtain the i-th road ultrasonic signal, and are carried out according to operational order calculating and obtaining signal ephemeral data,
And send signal ephemeral data to temporal cache space;
S3, obtains the ephemeral data based on the i-th obtained tunnels of step S2, with reference to the source ultrasonic signal on the i+1 road of acquisition, and
Computing is carried out according to operational order and signal results data are obtained, and then signal results data are sent to target cache space;
S4, circulation performs step S2, S3, and the signal results data that circulation obtains in step S3 every time are the echo signal on the i-th tunnel,
Until after all signal ways of traversal loop, all obtained echo signals are carried out into imaging.
2. the method that GPU according to claim 1 realizes ultrasonic signal filtering process, it is characterised in that the step S1
Also include:
It is the consistent multiplex ultrasonic signal of length, the total way of ultrasonic signal therein that ultrasonic signal is carried out into preliminary exposition
I value is less than N-1 to N between 64 and 256, and wherein.
3. the method that GPU according to claim 1 realizes ultrasonic signal filtering process, it is characterised in that the step S1
Also include:
Corresponding operational order is created according to different ultrasonic signal sources, and can be the combination of nonidentity operation instruction.
4. the method that GPU according to claim 1 realizes ultrasonic signal filtering process, it is characterised in that the step S2
Also include with S3:
The echo signal form of output therein is used to be consistent with input ultrasonic signal form.
5. the method that GPU according to claim 1 realizes ultrasonic signal filtering process, it is characterised in that the step S2
Also include with S3:
The ultrasonic signal obtained from former spatial cache is passed through circular treatment by GPU processing cores when obtaining ultrasonic signal
Obtained echo signal is distributed to the specified coordinate in target cache space, and wherein acquired ultrasonic signal coordinate can be certainly
Definition is set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610586805.8A CN107292809B (en) | 2016-07-22 | 2016-07-22 | Method for realizing ultrasonic signal filtering processing by GPU |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610586805.8A CN107292809B (en) | 2016-07-22 | 2016-07-22 | Method for realizing ultrasonic signal filtering processing by GPU |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107292809A true CN107292809A (en) | 2017-10-24 |
CN107292809B CN107292809B (en) | 2020-10-09 |
Family
ID=60093366
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610586805.8A Active CN107292809B (en) | 2016-07-22 | 2016-07-22 | Method for realizing ultrasonic signal filtering processing by GPU |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107292809B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060080508A1 (en) * | 2004-10-08 | 2006-04-13 | International Business Machines Corporation | Snoop filter directory mechanism in coherency shared memory system |
CN103544682A (en) * | 2013-09-17 | 2014-01-29 | 华中科技大学 | Non-local mean filter method for three-dimensional ultrasonic images |
CN103745447A (en) * | 2014-02-17 | 2014-04-23 | 东南大学 | Fast parallel achieving method for non-local average filtering |
CN104237859A (en) * | 2014-08-27 | 2014-12-24 | 武汉大学 | Method for achieving external illuminator radar multi-channel time domain clutter suppression by means of GPU |
-
2016
- 2016-07-22 CN CN201610586805.8A patent/CN107292809B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060080508A1 (en) * | 2004-10-08 | 2006-04-13 | International Business Machines Corporation | Snoop filter directory mechanism in coherency shared memory system |
CN103544682A (en) * | 2013-09-17 | 2014-01-29 | 华中科技大学 | Non-local mean filter method for three-dimensional ultrasonic images |
CN103745447A (en) * | 2014-02-17 | 2014-04-23 | 东南大学 | Fast parallel achieving method for non-local average filtering |
CN104237859A (en) * | 2014-08-27 | 2014-12-24 | 武汉大学 | Method for achieving external illuminator radar multi-channel time domain clutter suppression by means of GPU |
Also Published As
Publication number | Publication date |
---|---|
CN107292809B (en) | 2020-10-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20220261615A1 (en) | Neural network devices and methods of operating the same | |
WO2018000724A1 (en) | Cdvs extraction process acceleration method based on gpgpu platform | |
US10891779B2 (en) | Efficient volumetric reconstruction with depth sensors | |
CA2945022A1 (en) | Voxelization of a mesh | |
JP7414930B2 (en) | Information processing device, information processing method | |
CN107437273A (en) | Six degree of freedom three-dimensional reconstruction method, system and the portable terminal of a kind of virtual reality | |
CN109117950A (en) | The sparse tensor compression method of layering based on artificial intelligence equipment | |
Singhal et al. | Implementation and optimization of image processing algorithms on embedded GPU | |
TW201618037A (en) | Method and apparatus for efficient texture compression | |
Chen et al. | LANMC: LSTM-assisted non-rigid motion correction on FPGA for calcium image stabilization | |
CN107292809A (en) | A kind of method that GPU realizes ultrasonic signal filtering process | |
JP2006507051A5 (en) | ||
Binotto et al. | A cpu, gpu, fpga system for x-ray image processing using high-speed scientific cameras | |
US11664818B2 (en) | Neural network processor for compressing featuremap data and computing system including the same | |
CN115147283A (en) | Image reconstruction method, device, equipment and medium | |
WO2016048176A1 (en) | Method and apparatus for filtering compressed textures | |
JP5263854B1 (en) | Image compression apparatus, image decompression apparatus, image compression program, image decompression program, and data structure | |
KR20200056898A (en) | Processing apparatus and method for processing add operation thereof | |
Cadena et al. | Processing MRI brain image using OpenMP and fast filters for noise reduction | |
CN105991959B (en) | A kind of image processing method and electronic equipment | |
TWI789071B (en) | Image processing system and related image processing method for image enhancement based on region control and texture synthesis | |
Hiware et al. | Efficient Implementation of Selective Image Encryption Technique on Multi-core Reconfigurable System | |
KR20150022299A (en) | Method and apparatus for transforming coefficient information | |
JP6158576B2 (en) | Image compression apparatus, image decompression apparatus, image compression program, image decompression program, and data structure | |
CN115168794B (en) | Frequency spectrum analysis method and system based on improved DFT (discrete Fourier transform) and electronic equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |