CN108288247A - A kind of realtime graphic detail enhancing method on multi-core DSP platform - Google Patents
A kind of realtime graphic detail enhancing method on multi-core DSP platform Download PDFInfo
- Publication number
- CN108288247A CN108288247A CN201711293987.0A CN201711293987A CN108288247A CN 108288247 A CN108288247 A CN 108288247A CN 201711293987 A CN201711293987 A CN 201711293987A CN 108288247 A CN108288247 A CN 108288247A
- Authority
- CN
- China
- Prior art keywords
- core
- image
- ipc
- main
- completion
- 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
- 238000000034 method Methods 0.000 title claims abstract description 18
- 230000002708 enhancing effect Effects 0.000 title claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000001914 filtration Methods 0.000 claims abstract description 21
- 230000000694 effects Effects 0.000 claims abstract description 14
- 230000002146 bilateral effect Effects 0.000 claims abstract description 13
- 230000000295 complement effect Effects 0.000 claims abstract description 12
- 230000004927 fusion Effects 0.000 claims abstract description 5
- 230000004913 activation Effects 0.000 claims abstract description 4
- 238000012546 transfer Methods 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 2
- 238000005457 optimization Methods 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20028—Bilateral filtering
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to airborne graph and image processing field, a kind of be related on multi-core DSP platform image detail enhancement method.The present invention is based on single order cause and effect iteration two-sided filters to carry out details enhancing processing to image, using the parallel processing capability of multi-core DSP to algorithm for image enhancement parallel processing and optimization.Main core sends IPC and interrupts to from core, and the parallel filtering for being iterated bilateral filtering line direction to image after core activation is handled；After line direction has been filtered, each core sends IPC and interrupts has shown completion to main core.The subsequent supervention of complement mark that main core receives each core send IPC to interrupt to from core, and the parallel filtering processing that bilateral filtering column direction is iterated after interrupting is received from core, and being sent completely mark to main core after the completion of processing interrupts.After main core receives filtering complement mark, details enhancing and fusion display output are carried out.The present invention substantially increases the processing speed of image enhancement, and the significant effect of image enhancement can be applied to the higher field of requirement of real-time.
Description
Technical field
The invention belongs to airborne graph and image processing field, it is related to the image detail enhancing side on a kind of multi-core DSP platform
Method.
Background technology
Cockpit display system needs to provide from the required various information of overall processes such as take off, navigate, land.However low
It under the natural environments such as visibility or night, is had a certain difference between image and original image, the details letter in picture
Breath is often submerged in the background, can not be recognized, and is needed to optimize picture quality using image detail enhancing technology, is improved image
Clarity and interpretation property.Traditional image detail enhancement method complexity is higher, it is difficult to reach the requirement of real time implementation processing.
And due to the feature of traditional bilateral filtering non-iterative correlation, parallel optimization can not be carried out to it, again will to the optimization of algorithm itself
Cause the precision of algorithm to be difficult to ensure, cannot be satisfied the performance requirement of airborne display.Therefore, traditional image detail enhances algorithm
It is less in occasion (especially built-in field) application being strict with to image processing time.Need searching one kind that can meet
The performance requirement of image detail enhancing, and the image detail enhancement method of real-time processing requirement can be reached.
Invention content
The purpose of the present invention is：
Solve the problems, such as the image detail enhancing under the natural environments such as low visibility or night.
Present invention solves the technical problem that being：The present invention proposes the image detail enhancement method on a kind of multi-core DSP platform,
It is image layered using the New Iterative bilateral filtering progress that complexity is 0 (1), and at the multi-core parallel concurrent for making full use of DSP powerful
Reason ability, intercore communication are interrupted using IPC.The processing speed of image enhancement is substantially increased, the significant effect of image enhancement is full
The foot processing requirement of real-time.
The technical scheme is that：
The RGB image of the first step, main verification input is acquired, and is sent in IPC successively from core (setting from nuclear volume as N)
It is disconnected, it activates from core, while image is divided into N blocks (if cannot divide equally, the line number of each core distribution can be inconsistent, ensures simultaneously
Row execute optimal case), each image block with each from core correspond.N number of color space that executed parallel from core turns
Image is transformed into the spaces HSI by exchange the letters number by rgb space.IPC is sent after the completion to interrupt to main core.
Second step, main core receive complement mark interruption, extract I component image.Main core send out successively IPC interrupt to from
Core is activated from core；Meanwhile I component image is divided into N blocks (if cannot divide equally, method is same as above) according to line direction, and will be every
One image block corresponds to each from core.N number of single order cause and effect iteration bilateral filtering for executing line direction parallel from core, after the completion
Each core sends IPC and interrupts has shown completion to main core.
Third walks, main core receive it is N number of retransmit IPC from the complement mark of core and interrupt to respectively from core, equally by I component figure
It corresponds to each from core as being divided into N blocks (if cannot divide equally, method is same as above) according to column direction, respectively executes row parallel from core
The single order cause and effect iteration bilateral filtering in direction, each core sends IPC and interrupt to main core after the completion has shown completion.
4th step, main core receive N number of filtering complement mark from core, obtain the basal layer of I component image, and main core passes through
The levels of detail of I component image is calculated, processing is amplified to levels of detail.Main core will treated levels of detail and basal layer into
Row linear fusion obtains complete I component image；And merge with H, S component, obtain enhanced HSI images.Then it sends out successively
IPC is interrupted to N number of from core.
5th step, it is N number of from core receive interruption after execute space transfer function parallel, HSI images are converted into RGB image.
Hair IPC is interrupted to main core after the completion of each core executes.
6th step exports the enhanced image of details after main core activation.
As a kind of improvement of the technical program, the processing data and process data of image enhancement are all cached, and are protected
Buffer consistency is held, the acceleration effect for reading and writing data is notable.
Intercore communication is interrupted with synchronous using IPC, and occupancy resource is less, and multi-core parallel concurrent acceleration effect is notable.
More complex space transfer function is executed parallel by multinuclear, substantially increases the processing effect of entire algorithm
Rate.
As a kind of improvement of the technical program, carried out using novel single order cause and effect iteration two-sided filter image layered.
For each pixel, output y [n] is only related to current input x [n] and upper output y [n-1] of a pixel, i.e. y
[n]=(1-a) * x [n]+a*y [n-1], wherein a are feedback constant.Complexity is only 0 (1), and the output of each pixel is only
Weighted sum twice does not need any divide operations, calculates simply, is easy to parallel piecemeal processing, obvious processing effect.And it is traditional
Bilateral filtering it is more complicated, be related to complicated filter window movement and multiplication and division operation, complexity is 0 (w2), w is spectral window
The mouth length of side.
Beneficial effects of the present invention：The present invention is based on single order cause and effect iteration two-sided filters to carry out at details enhancing image
Reason protrudes the important information in image, improves the clarity and perceptibility of image, ensure that the detailed information of image is undistorted, schemes
The significant effect of image intensifying.Simultaneously as incoherent independently of each other between row and row, column and row in iteration bilateral filtering image
Feature carries out parallel processing and optimization to algorithm for image enhancement, substantially increases figure using the parallel processing capability of multi-core DSP
The speed of image intensifying algorithm can be applied to the higher field of requirement of real-time.
Description of the drawings
Fig. 1, which is image detail, enhances algorithm schematic diagram；
Fig. 2 is the functional block diagram that multi-core DSP platform executes.
Specific implementation mode
As shown in Fig. 2, by taking certain 8 core of type (Core0~Core7) DSP embedded system as an example, Core0 is main core, is responsible for more
The scheduling and pool of core；Core1~Core7 is to be responsible for executing the multi-core parallel concurrent task of Core0 distribution from core.Core0 is to input
Image be acquired, then by image color space conversion and iteration bilateral filtering (comprising line direction filtering and column direction
Filtering) task distributes to Core1~Core7 and executes parallel, filtering to Core0 feeds back complement mark after completing, and Core0 passes through
Are obtained by the enhanced image of details and is exported for levels of detail amplification, fusion calculation.
The technical program is described in further details below in conjunction with the accompanying drawings, equally with certain 8 core of type (Core0~Core7)
For DSP embedded systems.
The first step, Core0 are acquired the RGB image of input, are sent in IPC successively to from core Core1~Core7
It is disconnected, it activates from core, while image is divided into 7 pieces (if cannot divide equally, the line number of each core distribution can be inconsistent, ensures simultaneously
Row execute optimal case), each image block with each from core correspond.Core1~Core7 executes color sky parallel
Between transfer function, image is transformed into the spaces HSI by rgb space.IPC is sent after the completion to interrupt to Core0.
Second step, Core0 receive complement mark interruption, extract I component image.Core0 send out successively IPC interrupt to
Core1~Core7 is activated from core；Meanwhile I component image according to line direction is divided into 7 pieces (if cannot divide equally, method is same
On), and each image block corresponded to each from core.Core1~Core7 executes the bilateral filter of iteration of line direction parallel
Wave, each core sends IPC and interrupts after the completion has shown completion to Core0.
Third walks, and the complement mark that Core0 receives Core1~Core7 retransmits IPC and interrupts to respectively from core, equally by I
Component image according to column direction be divided into 7 pieces (if cannot divide equally, method is same as above) correspond to it is each from core, Core1~
Core7 executes the iteration bilateral filtering of column direction parallel, and each core sends IPC and interrupt to Core0 after the completion has shown completion.
4th step, Core0 receive the filtering complement mark of Core1~Core7, obtain the basal layer of I component image,
The levels of detail of I component image is obtained by calculation in Core0, and processing is amplified to levels of detail.Core0 will treated levels of detail
Linear fusion is carried out with basal layer, obtains complete I component image；And merge with H, S component, obtain enhanced HSI images.
Then hair IPC is interrupted to Core1~Core7 successively.
5th step, Core1~Core7 execute space transfer function after receiving interruption, HSI images are converted into RGB parallel
Image.Hair IPC is interrupted to Core0 after the completion of each core executes.
6th step exports the enhanced image of details after Core0 activation.
For the optimization part in DSP platform, by calling DSPLIB development libraries that can quickly execute certain instructions.Together
When, before algorithm for image enhancement processing, by the memory configurations of each core at Cache, and Cache consistency is kept, accelerates memory
The speed of access.
Claims (2)
1. the realtime graphic detail enhancing method on a kind of multi-core DSP platform, which is characterized in that include the following steps：
The RGB image of the first step, main verification input is acquired, if being N from nuclear volume, is interrupted, is swashed to IPC is sent successively from core
Living to be divided into N blocks from core, while by image, if cannot divide equally, the line number of each core distribution can be inconsistent, ensures parallel execute
Optimal case, each image block with each from core correspond；It is N number of to execute color space transfer function parallel from core,
Image is transformed into the spaces HSI by rgb space；IPC is sent after the completion to interrupt to main core；
Second step, main core receive complement mark interruption, extract I component image；Main core is sent out IPC and is interrupted to from core successively, swashs
It lives from core；Meanwhile I component image is divided into N blocks according to line direction, if cannot divide equally, method is same as above, and by each figure
As block correspond to it is each from core；N number of single order cause and effect iteration bilateral filtering for executing line direction parallel from core, after the completion each core hair
It send IPC to interrupt and has shown completion to main core；
Third walks, and main core, which receives, N number of to be retransmited IPC from the complement mark of core and interrupt to respectively from core, equally pressing I component image
N blocks are divided into according to column direction, if cannot divide equally, method is same as above, and corresponds to each from core, respectively executes column direction parallel from core
Single order cause and effect iteration bilateral filtering, after the completion each core to main core send IPC interruption has shown completion；
4th step, main core receive N number of filtering complement mark from core, obtain the basal layer of I component image, and main core passes through calculating
The levels of detail of I component image is obtained, processing is amplified to levels of detail；Main core will treated levels of detail and basal layer into line
Property fusion, obtain complete I component image；And merge with H, S component, obtain enhanced HSI images；Then IPC is sent out successively
It interrupts to N number of from core；
5th step, it is N number of from core receive interruption after execute space transfer function parallel, HSI images are converted into RGB image；Each core
Hair IPC is interrupted to main core after the completion of executing；
6th step exports the enhanced image of details after main core activation.
2. the realtime graphic detail enhancing method on a kind of multi-core DSP platform according to claim 1, it is characterised in that：
Further include that the processing data of image enhancement and process data are all cached, and keep buffer consistency.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711293987.0A CN108288247B (en) | 2017-12-07 | 2017-12-07 | Real-time image detail enhancement method on multi-core DSP platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711293987.0A CN108288247B (en) | 2017-12-07 | 2017-12-07 | Real-time image detail enhancement method on multi-core DSP platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108288247A true CN108288247A (en) | 2018-07-17 |
| CN108288247B CN108288247B (en) | 2021-04-20 |
Family
ID=62831889
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711293987.0A Active CN108288247B (en) | 2017-12-07 | 2017-12-07 | Real-time image detail enhancement method on multi-core DSP platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108288247B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109584143A (en) * | 2018-12-07 | 2019-04-05 | 中国科学院长春光学精密机械与物理研究所 | A kind of aerial camera image-enhancing equipment and method |
| CN111327356A (en) * | 2020-02-26 | 2020-06-23 | 航天恒星科技有限公司 | Demodulation device |
| CN112215785A (en) * | 2020-11-05 | 2021-01-12 | 中国航空工业集团公司西安航空计算技术研究所 | Infrared image detail enhancement method based on layered compression |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101546277A (en) * | 2009-04-27 | 2009-09-30 | 华为技术有限公司 | Multiple core processor platform and multiple core processor synchronization method |
| US20110242129A1 (en) * | 2010-04-02 | 2011-10-06 | Jianping Zhou | System, method and apparatus for an edge-preserving smooth filter for low power architecture |
| CN102270189A (en) * | 2011-06-17 | 2011-12-07 | 西安电子科技大学 | Inter-core communication method based on FPGA (Field Programmable Gate Array) multi-core system |
| CN105488769A (en) * | 2015-12-08 | 2016-04-13 | 中国航空工业集团公司西安航空计算技术研究所 | Real time video defogging method |
-
2017
- 2017-12-07 CN CN201711293987.0A patent/CN108288247B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101546277A (en) * | 2009-04-27 | 2009-09-30 | 华为技术有限公司 | Multiple core processor platform and multiple core processor synchronization method |
| US20110242129A1 (en) * | 2010-04-02 | 2011-10-06 | Jianping Zhou | System, method and apparatus for an edge-preserving smooth filter for low power architecture |
| CN102270189A (en) * | 2011-06-17 | 2011-12-07 | 西安电子科技大学 | Inter-core communication method based on FPGA (Field Programmable Gate Array) multi-core system |
| CN105488769A (en) * | 2015-12-08 | 2016-04-13 | 中国航空工业集团公司西安航空计算技术研究所 | Real time video defogging method |
Non-Patent Citations (4)
| Title |
|---|
| 朱才高: "红外图像增强算法研究及其DSP实现", 《中国优秀硕士学位论文全文数据库》 * |
| 王闯 等: "一种多核实时图像处理模块设计与实现", 《航空计算技术》 * |
| 白林亭 等: "多核DSP图像去雾优化方法", 《电光与控制》 * |
| 韩龙: "低能见度图像清晰化技术研究与DSP实现", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109584143A (en) * | 2018-12-07 | 2019-04-05 | 中国科学院长春光学精密机械与物理研究所 | A kind of aerial camera image-enhancing equipment and method |
| CN111327356A (en) * | 2020-02-26 | 2020-06-23 | 航天恒星科技有限公司 | Demodulation device |
| CN112215785A (en) * | 2020-11-05 | 2021-01-12 | 中国航空工业集团公司西安航空计算技术研究所 | Infrared image detail enhancement method based on layered compression |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108288247B (en) | 2021-04-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2016560B1 (en) | Advanced anti-aliasing with multiple graphics processing units | |
| CN111062858B (en) | Efficient rendering-ahead method, device and computer storage medium | |
| US11496773B2 (en) | Using residual video data resulting from a compression of original video data to improve a decompression of the original video data | |
| DE102020124932A1 (en) | Apparatus and method for real-time graphics processing using local and cloud-based graphics processing resources | |
| US20040010622A1 (en) | Method and system for buffering image updates in a remote application | |
| US10964000B2 (en) | Techniques for reducing noise in video | |
| CN106095437B (en) | The implementation method and device of the user interface layout type of RTL from right to left | |
| CN109785417B (en) | Method and device for realizing OpenGL cumulative operation | |
| CN105528207A (en) | Virtual reality system, and method and apparatus for displaying Android application images therein | |
| CN108288247A (en) | A kind of realtime graphic detail enhancing method on multi-core DSP platform | |
| CN110046116B (en) | Tensor filling method, device, equipment and storage medium | |
| KR102278021B1 (en) | Program code transformation to improve image processor runtime efficiency | |
| EP2662837A2 (en) | Real-time order-independent transparent rendering | |
| US20220036632A1 (en) | Post-processing in a memory-system efficient manner | |
| US20100013843A1 (en) | Method For Performing Undo And Redo Operations On A Graphics Processing Unit | |
| CN106055294A (en) | Layer composition optimization method and apparatus | |
| DE102020108526A1 (en) | ADAPTIVE PIXEL SCAN ORDER FOR DENSE RENDERING | |
| WO2004040514A1 (en) | Method of background colour removal for porter and duff compositing | |
| DE102021127982A1 (en) | STREAMING A FIELD OF LIGHT WITH LOSSLESS OR LOSSY COMPRESSION | |
| CN101321235A (en) | Image processing apparatus, image processing method | |
| CN111158840B (en) | Image carousel method and device | |
| CN108243355A (en) | Set-top box users interface loading method and system | |
| CN102306097B (en) | Method for performing real-time image processing on scene images in MultiGen-Vega | |
| CN103593822B (en) | The method and apparatus that frosted special effect processing is carried out to data image | |
| CN115049572A (en) | Image processing method, image processing device, electronic equipment and computer readable storage medium |
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 |