CN104318511B - A kind of computer display card and its image processing method - Google Patents
A kind of computer display card and its image processing method Download PDFInfo
- Publication number
- CN104318511B CN104318511B CN201410572961.XA CN201410572961A CN104318511B CN 104318511 B CN104318511 B CN 104318511B CN 201410572961 A CN201410572961 A CN 201410572961A CN 104318511 B CN104318511 B CN 104318511B
- Authority
- CN
- China
- Prior art keywords
- data
- buffering area
- image
- filtering module
- gpu
- 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.)
- Active
Links
Abstract
The invention discloses a kind of computer display cards, including the first filtering module, the second filtering module, third filtering module, and first buffering area, second buffering area and the third buffering area being connected respectively with the first filtering module, the second filtering module, third filtering module, first buffering area and second buffering area are commonly connected to the first GPU, first GPU and third buffering area are commonly connected to the 2nd GPU, 2nd GPU is connected to the 3rd GPU, and the 3rd GPU is connected to the 4th buffering area.The invention also discloses a kind of methods carrying out image procossing using above-mentioned computer display card.The present invention can improve the deficiencies in the prior art, by improving video card structure and processing method, under the premise of using same hardware, improve video card image-capable.
Description
Technical field
The present invention relates to image procossing display technology field, especially a kind of computer display card and its image processing method.
Background technology
With the development of technology, computer quickly enhances the processing capacity of image.Computer is electricity to the core of image procossing
Brain video card.In order to achieve the purpose that improve computer display card processing capacity, the prior art be all by increase video memory quantity or
It improves video card arithmetic speed and realizes.It is this that there are costs to increase more, hardware by way of promoting hardware performance simply
Technical bottleneck directly affects the problem of video card processing capacity.
Invention content
The technical problem to be solved in the present invention is to provide a kind of computer display card and its image processing methods, can solve existing
The deficiency of technology under the premise of using same hardware, improves video card image procossing by improving video card structure and processing method
Ability.
In order to solve the above technical problems, the technical solution used in the present invention is as follows.
A kind of computer display card, including the first filtering module, the second filtering module, third filtering module, and respectively with
One filtering module, the second filtering module, third filtering module connected first buffering area, second buffering area and third buffering area,
First buffering area and second buffering area are commonly connected to the first GPU, and the first GPU and third buffering area are commonly connected to the 2nd GPU,
2nd GPU is connected to the 3rd GPU, and the 3rd GPU is connected to the 4th buffering area.
Preferably, the first buffering area, second buffering area are identical with the saving format of third buffering area, with data
The form of block is stored using storehouse format, and there are one signatures for each data block setting.
Preferably, the saving format of the 4th buffering area is matrix format, lead between the 4th buffering area and the 3rd GPU
Cross two-way read-write access connection.
A kind of image processing method using above-mentioned computer display card includes the following steps：
A, image data passes through the first filtering module, the second filtering module and third filtering module, the first filter module successively
Structural characterization data in image data is stored into first buffering area by block, and the second filtering module is special by the color in image data
Sign data storage enters second buffering area, and the rendering characteristic in image data is stored into third and buffered by third filtering module
Area；
B, structural characterization data and color characteristic data are carried out integration processing by the first GPU, form the first intermediate of image；
C, the 2nd GPU is entirely handled the first intermediate of image with characteristic is rendered, and is formed among image second
Body；
D, the second intermediate of image is homogenized processing by the 3rd GPU, then carries out edge repair process.
Preferably, in step A, between the structural characterization data and color characteristic data of same image block and color
It is provided with unique position coordinates contacts list between characteristic and rendering characteristic.
Preferably, in step B, structural characterization data is combined first, forms complete image block；Then by face
Color characteristic data are replaced image structural characterization data in the block according to position coordinates contacts list, each color characteristic data
The coverage area of packet is more than corresponding structural characterization data packet；Finally, by the color of two color characteristic datas of lap
Degree and gray scale are weighted average treatment.
Preferably, in step C, it is special to color according to the mark position of position coordinates contacts list using characteristic is rendered
Sign data are rendered；In render process, several color characteristic datas are randomly selected first and are rendered, then with wash with watercolours
The position of the dyeing characteristic of dye is the center of circle, is gradually rendered outward, until whole color data renderings finish.
Preferably, in step D, the step of being homogenized processing to the second intermediate of image includes,
D11, the second intermediate of image is progressively scanned according to pixel, detecting distance and detection threshold value is set separately,
The position that the coloration of pixel in detecting distance or gray scale that scan are more than to detection threshold value is marked；
D12, corresponding original structural characterization data and color characteristic data on the position of label are re-started into extraction,
Then the position distribution that color structure feature is determined according to structured data feature, by color characteristic data packet adjacent on position into
Row equalization carries out integration processing with structured data feature using the color characteristic data after equalization, is formed in homogenizing first
Mesosome；
D13, it is weighted averagely using the position marked in the first intermediate of homogenizing and step D11.
It is using advantageous effect caused by above-mentioned technical proposal：The present invention, which changes, in the prior art carries out image
The disposable pattern directly handled, by the way that image data is divided into three parts, and carries out contact appropriate to it, reduces number
According to the repetitive operation in processing procedure, processing speed is improved.In addition, by using three layers of data structure, it can be further
The display quality of image is advanced optimized, this optimization is by being then based on above-mentioned data structure, so increased fortune
Calculation amount is minimum, and additional computational burden will not be brought to system.
Description of the drawings
Fig. 1 is the schematic diagram of hardware configuration in a specific implementation mode of the invention.
In figure：1, the first filtering module；2, the second filtering module；3, third filtering module；4, first buffering area；5, second
Buffering area；6, third buffering area；7, the first GPU；8, the 2nd GPU；9, the 3rd GPU；10, the 4th buffering area；11, two-way read-write is logical
Road.
Specific implementation mode
The standardized element used in the present invention can commercially, and shaped piece is according to specification and attached drawing
Record can carry out customized, and the specific connection type of each part is all made of bolt ripe in the prior art, rivet, weldering
The conventional means such as connect, paste, this will not be detailed here.
Referring to Fig.1, one specific implementation mode of the present invention includes the first filtering module 1, the second filtering module 2, third mistake
Filter module 3, and be connected respectively with the first filtering module 1, the second filtering module 2, third filtering module 3 first buffering area 4,
Second buffering area 5 and third buffering area 6, first buffering area 4 and second buffering area 5 are commonly connected to the first GPU7, the first GPU7
It is commonly connected to the 2nd GPU8 with third buffering area 6, the 2nd GPU8 is connected to the 3rd GPU9, and the 3rd GPU9 is connected to the 4th buffering
Area 10.
It is worth noting that, the first buffering area 4, second buffering area 5 are identical with the saving format of third buffering area 6,
It is stored using storehouse format in the form of data block, there are one signatures for each data block setting.
It is worth noting that, the saving format of the 4th buffering area 10 is matrix format, the 4th buffering area 10 and third
It is connected by two-way read-write access 11 between GPU9.
A kind of image processing method using above-mentioned computer display card includes the following steps：
A, image data passes through the first filtering module 1, the second filtering module 2 and third filtering module 3, the first filtering successively
Structural characterization data in image data is stored into first buffering area 4 by module 1, and the second filtering module 2 will be in image data
Color characteristic data is stored into second buffering area 5, and the rendering characteristic in image data is stored by third filtering module 3
Three buffering areas 6；
B, structural characterization data and color characteristic data are carried out integration processing by the first GPU7, are formed among image first
Body；
C, the 2nd GPU8 is entirely handled the first intermediate of image with characteristic is rendered, and is formed among image second
Body；
D, the second intermediate of image is homogenized processing by the 3rd GPU9, then carries out edge repair process.
In step A, between the structural characterization data and color characteristic data of same image block and color characteristic data and
It is provided with unique position coordinates contacts list between rendering characteristic.
In step B, structural characterization data is combined first, forms complete image block；Then by color characteristic number
Image structural characterization data in the block is replaced according to according to position coordinates contacts list, the covering of each color characteristic data packet
Range is more than corresponding structural characterization data packet；Finally, by the coloration and gray scale of two color characteristic datas of lap
It is weighted average treatment, wherein coloration accounts for 70% weight, and gray scale accounts for 30% weight.
In step C, using render characteristic according to position coordinates contacts list mark position to color characteristic data into
Row renders；In render process, several color characteristic datas are randomly selected first and are rendered, then with rendered dyeing
The position of feature is the center of circle, is gradually rendered outward, until whole color data renderings finish.
In step D, the step of being homogenized processing to the second intermediate of image includes,
D11, the second intermediate of image is progressively scanned according to pixel, detecting distance and detection threshold value is set separately,
The position that the coloration of pixel in detecting distance or gray scale that scan are more than to detection threshold value is marked；Wherein, it detects
Distance is 10 pixel distances, and detection threshold value is that the ratio of two extreme values of size is 2: 1；
D12, corresponding original structural characterization data and color characteristic data on the position of label are re-started into extraction,
Then the position distribution that color structure feature is determined according to structured data feature, by color characteristic data packet adjacent on position into
Row equalization carries out integration processing with structured data feature using the color characteristic data after equalization, is formed in homogenizing first
Mesosome；
D13, it is weighted averagely using the position marked in the first intermediate of homogenizing and step D11, uses 1: 1 first
Ratio is weighted averagely, is then detected again, if still exceeding detection threshold value, according to variation tendency to weighting ratio
Example is adjusted, and is repeated the above process, until result is fallen within detection threshold value.
In step D, include to the step of the second intermediate of image progress edge repair process,
D21, sampling processing is carried out to image border, each structural characterization data packet positioned at edge is adopted including at least one
Sampling point；
D22, the image of the second intermediate of the first intermediate of image and image of sampled point is read out, to two images
The coloration and gray scale of point are compared, if consistency is all higher than 70%, are labeled as correct point, are otherwise labeled as erroneous point；
D23, carry out linear regression will be correctly put, regressive object equation is ax2+bx+(cx/ x)+d=y, wherein parameter a, b,
C, d is determined by linear regression algorithm.
D24, image border is recalculated using calculated equation of linear regression, obtains final result.
Finally, the 4th buffering area 10 exports handling result.
By the GTX650 video cards of NVIDIA brands available on the market and made of being encapsulated with GTX650 same hardwares
The video card of the present embodiment is compared, and comparison process uses same image and same image output quality as standard, and two aobvious
Block it is each carried out 100 experiments, the average treatment speed acquired is：
| Product | Average treatment speed (ns) |
| GTX650 | 3.5 |
| The present embodiment | 1.9 |
The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention.The technology of the industry
Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the above embodiments and description only describe this
The principle of invention, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes
Change and improvement all fall within the protetion scope of the claimed invention.The claimed scope of the invention by appended claims and its
Equivalent thereof.
Claims (8)
1. a kind of computer display card, it is characterised in that：Including the first filtering module (1), the second filtering module (2), third filter module
Block (3), and be connected respectively with the first filtering module (1), the second filtering module (2), third filtering module (3) first buffering
Area (4), second buffering area (5) and third buffering area (6), first buffering area (4) and second buffering area (5) are commonly connected to first
GPU (7), the first GPU (7) and third buffering area (6) are commonly connected to the 2nd GPU (8), and the 2nd GPU (8) is connected to the 3rd GPU
(9), the 3rd GPU (9) is connected to the 4th buffering area (10)；
Wherein, when handling image data, image data passes through the first filtering module (1), the second filtering module successively
(2) and the structural characterization data in image data can be stored into first by third filtering module (3), the first filtering module (1)
Color characteristic data in image data can be stored into second buffering area (5) by buffering area (4), the second filtering module (2), the
Rendering characteristic in image data can be stored into third buffering area (6) by three filtering modules (3).
2. computer display card according to claim 1, it is characterised in that：The first buffering area (4), second buffering area (5)
It is identical with the storage format of third buffering area (6), it is stored using storehouse format in the form of data block, each data block setting
There are one signatures.
3. computer display card according to claim 1, it is characterised in that：The storage format of 4th buffering area (10) is square
Grid array formula is connect between the 4th buffering area (10) and the 3rd GPU (9) by two-way read-write access (11).
4. a kind of image processing method using computer display card according to any one of claims 1 to 3, it is characterised in that packet
Include following steps：
A, image data passes through the first filtering module (1), the second filtering module (2) and third filtering module (3), the first mistake successively
Structural characterization data in image data is stored into first buffering area (4) by filter module (1), and the second filtering module (2) is by image
Color characteristic data in data is stored into second buffering area (5), and third filtering module (3) is by the rendering feature in image data
Data are stored into third buffering area (6)；
B, structural characterization data and color characteristic data are carried out integration processing by the first GPU (7), form the first intermediate of image；
C, the first intermediate of image is carried out integration processing by the 2nd GPU (8) with characteristic is rendered, and is formed among image second
Body；
D, the second intermediate of image is homogenized processing by the 3rd GPU (9), then carries out edge repair process.
5. image processing method according to claim 4, it is characterised in that：In step A, the structure feature of same image block
It is provided with unique position coordinates between data and color characteristic data and between color characteristic data and rendering characteristic
Contacts list.
6. image processing method according to claim 5, it is characterised in that：In step B, first by structural characterization data into
Row combination, forms complete image block；Then by color characteristic data according to position coordinates contacts list to image structure in the block
Characteristic is replaced, and the coverage area of each color characteristic data packet is more than corresponding structural characterization data packet；Finally,
The coloration of two color characteristic datas of lap and gray scale are weighted average treatment.
7. image processing method according to claim 5, it is characterised in that：In step C, using render characteristic according to
The mark position of position coordinates contacts list renders color characteristic data；In render process, several are randomly selected first
Color characteristic data is rendered, and then using the position of rendered color characteristic as the center of circle, is gradually rendered outward, until complete
Portion's color data rendering finishes.
8. image processing method according to claim 4, it is characterised in that：In step D, to the second intermediate of image into
Row homogenizing processing the step of include,
D11, the second intermediate of image is progressively scanned according to pixel, detecting distance and detection threshold value is set separately, will sweep
The position that the coloration of pixel in detecting distance or gray scale described is more than detection threshold value is marked；
D12, corresponding original structural characterization data and color characteristic data on the position of label are re-started into extraction, then
The position distribution that color structure feature is determined according to structured data feature puts down color characteristic data packet adjacent on position
Homogenizing carries out integration processing with structured data feature using the color characteristic data after equalization, forms the first intermediate of homogenizing；
D13, it is weighted averagely using the position marked in the first intermediate of homogenizing and step D11.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410572961.XA CN104318511B (en) | 2014-10-24 | 2014-10-24 | A kind of computer display card and its image processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410572961.XA CN104318511B (en) | 2014-10-24 | 2014-10-24 | A kind of computer display card and its image processing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104318511A CN104318511A (en) | 2015-01-28 |
| CN104318511B true CN104318511B (en) | 2018-10-26 |
Family
ID=52373737
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410572961.XA Active CN104318511B (en) | 2014-10-24 | 2014-10-24 | A kind of computer display card and its image processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104318511B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105139356B (en) * | 2015-08-25 | 2018-06-22 | 北京锤子数码科技有限公司 | The frosted glass effect processing method and device of a kind of image data |
| CN107729095B (en) * | 2017-09-13 | 2020-12-04 | 深信服科技股份有限公司 | Image processing method, virtualization platform and computer-readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101542524A (en) * | 2006-11-08 | 2009-09-23 | 英特尔公司 | Graphics processor pipelined reduction operations |
| CN102109899A (en) * | 2009-12-23 | 2011-06-29 | 联想(北京)有限公司 | Computer, displayer, and display method of computer |
| CN103309958A (en) * | 2013-05-28 | 2013-09-18 | 中国人民大学 | OLAP star connection query optimizing method under CPU and GPU mixing framework |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103810124A (en) * | 2012-11-09 | 2014-05-21 | 辉达公司 | Data transmission system and data transmission method |
| US9104593B2 (en) * | 2012-12-14 | 2015-08-11 | Advanced Micro Devices, Inc. | Filtering requests for a translation lookaside buffer |
-
2014
- 2014-10-24 CN CN201410572961.XA patent/CN104318511B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101542524A (en) * | 2006-11-08 | 2009-09-23 | 英特尔公司 | Graphics processor pipelined reduction operations |
| CN102109899A (en) * | 2009-12-23 | 2011-06-29 | 联想(北京)有限公司 | Computer, displayer, and display method of computer |
| CN103309958A (en) * | 2013-05-28 | 2013-09-18 | 中国人民大学 | OLAP star connection query optimizing method under CPU and GPU mixing framework |
Non-Patent Citations (4)
| Title |
|---|
| "3D显卡术语简介";孔明;《计算机与网络》;20001212(第23期);12-13 * |
| "图像特征提取研究";翟俊海 等;《河北大学学报(自然科学版)》;20090125;第29卷(第1期);106-112 * |
| "基于通用可编程GPU的视频解码器--架构、算法与实现";房波;《中国优秀博硕士学位论文全文数据库（硕士） 信息科技辑》;20050615(第02期);I135-223 * |
| "显卡的结构和工作原理";洪伟;《电脑应用文萃》;20000318(第03期);11-13 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104318511A (en) | 2015-01-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110060237B (en) | Fault detection method, device, equipment and system | |
| CN107230202B (en) | Automatic identification method and system for road surface disease image | |
| CN103996170B (en) | Image edge saw-tooth eliminating method with super resolution | |
| CN103530878B (en) | A kind of edge extracting method based on convergence strategy | |
| CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
| CN110084241B (en) | Automatic ammeter reading method based on image recognition | |
| CN107014819A (en) | A kind of solar panel surface defects detection system and method | |
| CN103810474A (en) | Car plate detection method based on multiple feature and low rank matrix representation | |
| CN103593670A (en) | Copper sheet and strip surface defect detection method based on-line sequential extreme learning machine | |
| CN109800619B (en) | Image recognition method for citrus fruits in mature period | |
| CN109727233B (en) | LCD defect detection method | |
| CN104990925A (en) | Defect detecting method based on gradient multiple threshold value optimization | |
| CN106855951A (en) | A kind of grain kind quality detecting method based on computer vision | |
| CN104680130A (en) | Chinese character recognition method for identification cards | |
| CN105319223B (en) | The detection method and detecting system of a kind of chalk rice | |
| CN108955901B (en) | Infrared temperature measurement method and system and terminal equipment | |
| CN110335233B (en) | Highway guardrail plate defect detection system and method based on image processing technology | |
| CN105225216A (en) | Based on the Iris preprocessing algorithm of space apart from circle mark rim detection | |
| CN103268492B (en) | A kind of corn grain type identification method | |
| CN104318511B (en) | A kind of computer display card and its image processing method | |
| CN115266732A (en) | Carbon fiber tow defect detection method based on machine vision | |
| CN104112123A (en) | Defect characteristic extraction and identification method of AOI system used for bullet apparent defect detection | |
| CN110147792B (en) | Medicine package character high-speed detection system and method based on memory optimization | |
| CN108805854B (en) | Method for rapidly counting tablets and detecting completeness of tablets in complex environment | |
| CN206897873U (en) | A kind of image procossing and detecting system based on detection product performance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20190111 Address after: 343100 No. 192 Torch Avenue, Jinggangshan Economic Development Zone, Ji'an City, Jiangxi Province Patentee after: Jiangxi creation Microelectronics Co., Ltd. Address before: 343100 Jinggangshan Economic and Technological Development Zone, Ji'an, Jiangxi Province Patentee before: JIANGXI CHUANGCHENG ELECTRONIC CO., LTD. |