CN105023241A - Fast image interpolation method for mobile terminal - Google Patents
Fast image interpolation method for mobile terminal Download PDFInfo
- Publication number
- CN105023241A CN105023241A CN201510460897.0A CN201510460897A CN105023241A CN 105023241 A CN105023241 A CN 105023241A CN 201510460897 A CN201510460897 A CN 201510460897A CN 105023241 A CN105023241 A CN 105023241A
- Authority
- CN
- China
- Prior art keywords
- edge
- interpolation
- image
- algorithm
- information
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4023—Decimation- or insertion-based scaling, e.g. pixel or line decimation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4053—Super resolution, i.e. output image resolution higher than sensor resolution
Abstract
The present invention discloses a fast image interpolation method for a mobile terminal. The fast image interpolation method comprises the steps of: firstly, performing edge detection on a low-resolution original image to obtain edge information, calculating the strength of an edge according to the edge information of the low-resolution original image and a human visual system, expanding the edge according to the strength, dividing the image into an edge region and a non-edge region according to the expanded edge, processing the edge region by adopting an interpolation algorithm with relatively high fidelity, and storing the edge information; and secondly, processing the non-edge region by adopting a faster bicubic interpolation algorithm, then further sharpening the edge of an interpolation image according to the existing edge information to reduce the blurriness of the edge and prompt the visual quality of the image, and at last combining the interpolation algorithm with an NEON parallel technology to obtain a high-resolution image subjected to parallel optimization. The result of the application of the technology to a mobile video monitoring system shows that the technology can ensure that multimedia applications of a mobile phone can smoothly interpolate high-resolution images for playing.
Description
Technical field
The present invention relates to image interpolation technology and mobile terminal parallel calculating field, refer in particular to a kind of NEON technology that utilizes being applicable to mobile terminal and carry out the rapid image interpolation method of parallel processing optimization.
Background technology
Image interpolation is a kind of conventional image processing techniques in Digital Image Processing.Image interpolation is exactly by original low-resolution image, through the regeneration of view data, produces the process of high-resolution target image.In Interpolation Process, owing to only having signal on the sampled point in original low-resolution image to be known, but not on sampled point, signal is not known, thus inevitably there will be pseudomorphism.As phenomenons such as Block Artifact (BlockingArtifacts), ringing artifacts (Ring Artifacts), aliasing artifacts (Jaggies), especially edge region is more obvious.Because the essence of these pseudomorphisms is loss of edge detail information (high-frequency region).How to reduce the generation of this kind of pseudomorphism in Interpolation Process, be the problem being worth research as far as possible.
At present, if classified according to interpolation method, image interpolation algorithm can be divided into two large classes: linear interpolation and non-linear interpolation.Linear interpolation algorithm is that interpolation pixel coordinate in high resolving power target image is mapped to low resolution original image, according to original image known pixel values discrete around interpolation pixel, carry out interpolation by interpolating function (i.e. linear interpolation kernel function), thus calculate the pixel value of interpolation point on high resolving power target image.Linear interpolation algorithm, is normally weighted on average known pixel value, or perhaps image carries out the pixel value that two-dimensional convolution operation draws interpolation point between discrete sampled value and interpolating function.
The present invention be directed to a kind of image interpolation fast method being applicable to mobile terminal, therefore the time complexity of interpolation algorithm is a very important index for purposes of the invention.Conventional traditional interpolation algorithm has: arest neighbors interpolation algorithm (Nearest Neighbor), bilinear interpolation algorithm (Bilinear), bicubic interpolation algorithm (Bicubic), desirable Sinc function interpolation algorithm (Sinc), B-spline interpolation algorithm (B-Spline), Lanczos interpolation algorithm (Lanczos).The present invention will analyze these algorithms, and contrast their algorithm complex.Test pattern comprises Shapes (384 × 192), Lena (512 × 512), Wheel (320 × 180), as shown in accompanying drawing 1,2,3.
In order to test the time loss of various interpolation algorithm to high-definition image process, we carry out interpolation amplification experiment to Shapes image, choose different interpolation algorithm, respectively 4 times of interpolation amplifications are carried out to the original low-resolution image of 480 × 270, obtain the interpolation image of 1920 × 1080.The time of record algorithm consumption also tabulates and compares.Error in measuring and calculating, we carry out 3 groups of experiments to often kind of algorithm respectively, and often group comprises 10000 continuous print interpolation magnification operations, and statistics often group tests the time consumed, and calculates mean consumption time and average treatment speed.
The working time that accompanying drawing 4 reflects various different conventional interpolation algorithm contrasts situation.We can find, along with the kernel function of interpolation algorithm becomes increasingly complex, the time complexity of interpolation algorithm is also more and more higher, and the time that Interpolation Process consumes also will increase.For Lanczos serial algorithm, along with the increase of window size, the time that Lanczos algorithm consumes also sharply rises.
On mobile terminals, such as the application such as mobile video monitor, mobile video order is very responsive for the execution time of interpolation algorithm in multimedia application.On the one hand, if interpolation algorithm efficiency is not high, interpolation elapsed time is long, even if the clear quality of picture is better, also can bring bad Consumer's Experience owing to too blocking to time to user.On the other hand, the mobile terminal device low to flying power is brought more energy resource consumption by inefficient interpolation algorithm, causes its flying power to reduce, and equipment heating increases.Therefore, the interpolation algorithm research on mobile terminal, except will considering its subjective and objective quality, also must be placed on critical positions by its working time and consider.According to the data of each algorithm process Shapes image, consider interpolation quality (for Y-PSNR (PSNR) value) and the time loss of conventional traditional images interpolation algorithm, using these two factors as two dimensions, draw and obtain accompanying drawing 5.
As can be seen from accompanying drawing 5, algorithms of different has different interpolation objective qualities and different elapsed times.Tu Zhong elapsed time unit is second every ten thousand times.Data analysis from figure, the time that the algorithm that interpolation quality is better generally consumes is more.Arest neighbors interpolation algorithm time loss is very low, but its objective quality is also very poor.Sinc interpolation algorithm time loss is very high, but quality does not obviously promote.Comprehensive various traditional normal image interpolation algorithm, bicubic interpolation algorithm (Bicubic) has good balance between performance and cost.More complicated traditional interpolation algorithm (such as B-spline interpolation algorithm) consumes the more time, but increased quality not obvious.
Experiment shows, these three kinds of interpolation algorithm complexities of arest neighbors interpolation algorithm, bilinear interpolation algorithm, bicubic interpolation algorithm are lower, carries out realtime graphic amplifieroperation than being more suitable for mobile terminal.The new method that the present invention proposes attempts, in raising interpolation quality, to keep the high efficiency of algorithm while the fringe region interpolation quality that especially human eye is the most responsive.
Current most mobile terminal is all adopt traditional interpolation algorithm to carry out Nonlinear magnify.But traditional image interpolation algorithm, do not make full use of the prior imformation of image, identical interpolation kernel function process is all carried out to all parts of image, and the pixel only employed around interpolation point calculates, lack enough information, cause the phenomenons such as the appearance of the image border after interpolation is fuzzy, sawtooth.
By analyzing above numerous interpolation algorithm, can find that some algorithm computation complexity is low, but interpolation is of poor quality; The interpolation quality of some algorithm is good, but calculation cost is but that mobile terminal is unaffordable.The interpolation algorithm being applicable to mobile terminal should possess following characteristics:
The first, algorithm complex is low.Because embedded device computing power is relatively low, multimedia application requirement of real-time is high, is limited to battery durable ability simultaneously, and this just requires that the interpolation algorithm on embedded mobile terminal should have alap complexity.
The second, enlargement factor is high.At present, the network bandwidth is an important bottleneck factor of moving-limiting multimedia application development.In order to save the network bandwidth, the normally low-resolution image of the application program transmitting-receiving on mobile device, as QCIF (352*288).Now common mobile phone screen resolution has reached WVGA (960*640) or higher.When showing image, interpolation amplification certainly will to be carried out to adapt to screen true resolution to low-resolution image.This just causes mobile device needs to amplify the image of higher multiple, and this must cause the target image after amplifying comparatively fuzzy compared with source images.This just has higher requirement, as edge quality to the quality of interpolator arithmetic.
3rd, magnification ratio is variable arbitrarily.Amplify different from fixed proportion in traditional game, mobile device model is different, and screen resolution varies.Mobile solution needs adaptive different screen resolutions, just must support to carry out magnification ratio Nonlinear magnify variable arbitrarily.
When ensureing lower computation complexity, promote handset image interpolation quality as much as possible, be emphasis of the present invention, up to now there are no utilizing NEON technology to carry out the reported in literature of parallel optimization to interpolation algorithm at mobile terminal, the present invention has its originality.
Summary of the invention
The object of the invention is to overcome the shortcoming of prior art and deficiency, a kind of mobile terminal rapid image interpolation method is provided, when ensureing lower computation complexity, promoting handset image interpolation quality as much as possible, obtaining high-definition picture.
For achieving the above object, technical scheme provided by the present invention is: a kind of mobile terminal rapid image interpolation method, comprises the following steps:
1) rim detection is carried out to low resolution original image, obtain marginal information;
2) calculate the intensity at edge according to low-resolution image marginal information and human visual system, expand according to intensity edge;
3) according to the edge after expansion, image is divided into fringe region and non-edge, edge region adopts B-spline interpolation algorithm or SAI interpolation algorithm to carry out processing and preserving marginal information, adopts bicubic interpolation algorithm to process to non-edge;
4) according to existing marginal information, further Edge contrast is carried out to the edge of interpolation image, reduce marginal portion fog-level, the visual quality of prompting image;
5) interpolation algorithm is combined with NEON concurrent technique, to obtain the high-definition picture of parallel optimization process.
In step 2) in, first edge strength analysis is carried out according to low-resolution image, in edge strength information, except recording position in the picture, edge, also analyze according to human visual system's edge intensity, the power according to edge strength carries out extended operation to it.
In step 3) in; by the pixel-map in target image on original image; according to edge strength information by image zoning; for the flat site of non-edge; bicubic interpolation algorithm is adopted to carry out interpolation processing; adopt B-spline interpolation algorithm or SAI interpolation algorithm with Protect edge information information for fringe region; merge the interpolation result in each region; after obtaining high resolving power interpolation image; again further edge is done to its edge and strengthen process, reach algorithm execution time efficiency and the qualitative balance of algorithm interpolation.
In step 5) in, to computation complexity, high and algorithm the module of parallelization can carry out inline assembler optimization, automatic vectorization is adopted to utilize compiler to be optimized to other modules or function, adopt inline assembler that convolution algorithm, look are melted computing, the optimization of these computings of vector operation employing NEON assembly instruction, thus make this partial code can obtain parallel processing by SIMD framework, boosting algorithm execution efficiency.
Compared with prior art, tool has the following advantages and beneficial effect in the present invention:
The present invention is a kind of under lower time complexity, the handset image interpolation technique of parallel processing optimization is carried out in conjunction with NEON technology, be one by original low-resolution image, through the regeneration of view data, produce the process of high-resolution target image, handset image interpolation quality can be promoted as much as possible, obtain high-definition picture.Be proven, this technology result be applied in the middle of mobile video monitor system shows that it can make cellphone multimedia apply and can interpolation go out high-definition picture and play glibly, has that algorithm complex is low, enlargement factor is high, a magnification ratio feature such as variable arbitrarily.
Accompanying drawing explanation
Fig. 1 is one of test pattern selected.
Fig. 2 is the test pattern two selected.
Fig. 3 is the test pattern three selected.
The time comparison diagram that Fig. 4 consumes for 10000 times with different interpolation algorithm process for Shapes test pattern.
Fig. 5 is interpolation objective quality and elapsed time Integrated comparative figure.
Fig. 6 is the process flow diagram of mobile terminal rapid image interpolation method of the present invention.
Fig. 7 is the edge-protected algorithm process process flow diagram based on LR image.
Fig. 8 is the edge-protected algorithm process process flow diagram based on HR image.
Fig. 9 is to Wheel interpolation 10000 comparison diagrams consuming time on mobile phone.
Figure 10 is to Wheel interpolation 10000 Integrated comparative figure on mobile phone.
Embodiment
Below in conjunction with specific embodiment, the invention will be further described.
Propose in recent years much based on the image interpolation algorithm of marginal information.This kind of edge-protected algorithm is mainly divided into two classes: based on the algorithm of original LR image border and the algorithm based on HR image border after interpolation.
First edge-protected algorithm based on original low-resolution image (LR image) detects the edge of original low-resolution image, according to marginal information, image is divided into fringe region and the large class of non-edge two.Traditional interpolation algorithm is adopted for non-edge, special interpolation algorithm is adopted for fringe region, to reach the object of Protect edge information details.Its advantage is that interpolation speed is fast.Particular flow sheet as shown in Figure 7.
First edge-protected algorithm based on high-definition picture after interpolation (HR image) adopts traditional interpolation algorithm to carry out interpolation amplification to original low-resolution image; then the edge of the high-definition picture that interpolation obtains is detected; edge area pixel do special processing (as sharpening, remove fuzzy etc.), to reach the object strengthening edge details.Its advantage amplifies better quality, but speed is slower.This is because more than low-resolution image several times of the pixel quantity of high resolving power interpolation image, therefore edge analysis is done to high-definition picture and want comparison low-resolution image to do the many consumption of the edge analysis several-fold time.In addition, after obtaining high-definition picture marginal information, also need the cost edge of plenty of time to high-definition picture to do and remove fuzzy grade for special processing, this also will spend the more time.The process flow diagram of this method as shown in Figure 8.
The present invention, in conjunction with above-mentioned two kinds of advantages based on edge-protected interpolation algorithm, proposes a kind of new rapid image interpolation method being applicable to mobile terminal.This interpolation method is a kind of interpolation algorithm based on edge analysis, first edge strength analysis is carried out according to low-resolution image, in edge strength information, except recording position in the picture, edge, also analyze according to human visual system (HVS) edge intensity, the power according to edge strength carries out extended operation to it.Then by the pixel-map in target image on original image, according to edge strength information by image zoning.For the flat site of non-edge, adopt bicubic interpolation algorithm to carry out interpolation processing, adopt the interpolation algorithm of better quality with Protect edge information information (such as SAI interpolation or B-spline interpolation) for fringe region.Merge the interpolation result in each region, after obtaining high resolving power interpolation image, then the enhancing process of further edge is done to its edge, reach algorithm execution time efficiency and the qualitative balance of algorithm interpolation.Cellphone multimedia is applied can interpolation go out high-definition picture and play glibly.
As shown in Figure 6, the inventive method is divided into following a few step:
The first step: rim detection is carried out to low resolution original image, obtains marginal information.
Second step: the intensity calculating edge according to low-resolution image marginal information and human visual system, expands according to intensity edge.
3rd step: image is divided into fringe region and non-edge according to the edge after expansion, edge region adopts the higher interpolation algorithm of fidelity to carry out processing (such as B-spline interpolation algorithm or SAI interpolation algorithm) and preserving marginal information; The bicubic interpolation algorithm of speed is adopted to process to non-edge.
4th step: according to existing marginal information, carries out further Edge contrast to the edge of interpolation image, reduces marginal portion fog-level, the visual quality of prompting image.
5th step: be combined with NEON concurrent technique by interpolation algorithm, to obtain the high-definition picture of parallel optimization process.
Utilize ARM NEON general purpose single multiple instruction multiple data stream (MIMD) (Single Instruction Multiple Data) engine, the data of multimedia form can be processed efficiently.This has very great help to the lifting of Consumer's Experience.
NEON technology is mainly used in providing powerful to Multimedia Program and accelerating function flexibly.This realizes based on 128 single-instruction multiple-data stream (SIMD) frameworks expansion of arm processor.
Specifically, NEON provides the instruction set of a set of uniqueness, is characterized in single-instruction multiple-data stream (SIMD) instruction set.As its name suggests, this single-instruction multiple-data stream (SIMD) instruction set ability that an instruction cycle can be provided to operate multiple data.
During exploitation NEON instruction set program, use the assembly instruction under ARM platform.For Android mobile phone, adopt NDK (Native Development Kit) and Android SDK (Software Development Kit) JNI (JavaNative Interface) technology, can can perform storehouse .so file by using the algorithm of C/C++ language compilation to be packaged into, and loaded by Android SDK (Software Development Kit) JAVA language and call this library file.The benefit done like this is that C/C++ language is more suitable for doing some bottom arithmetic operations than JAVA language, for the CPU intensive procedure of the data processings such as image interpolation, adopts C/C++ language compilation to obtain better execution efficiency.
In addition, NEON assembly instruction collection can with C/C++ program shuffling.That is more efficient NEON parallel instruction can be embedded to reach the object of optimizer execution efficiency in C/C++ code.From soft project angle, compared with directly stiff ground Eembedded Assembly instruction in algorithm, more graceful and flexibly method be the algorithm function NEON assembly instruction realization that will commonly use, and be packaged into a function that directly can call with C/C++.
The present invention adopts inline assembler by conventional algorithm operating as convolution algorithm, look melt computing, vector operation etc. and adopt the optimization of NEON assembly instruction.Thus make this partial code can obtain parallel processing by SIMD framework, boosting algorithm execution efficiency.Meanwhile, remaining the encapsulation format of C/C++ owing to optimizing the NEON function of version, making C/C++ language can call NEON function as calling a common C/C++ function.By function pointer, Dynamic Selection can to call C/C++ version generic function or NEON version and line function, conveniently carry out contrast test.
The present invention adopts NEON to be optimized proposed algorithm.Consider that directly execution assembly code effect of optimization is better, in order to excavate NEON effect of optimization as far as possible, to computation complexity, higher and algorithm the module of parallelization can carry out inline assembler optimization in the present invention, adopts automatic vectorization to utilize compiler to be optimized to other modules or function.Consider that NEON assembly code can encapsulate function with the C/C++ in Function inlining and mutually change, and the latter has clear succinct feature, presents with the form of Inline Function the optimization of algorithm.
Real-time mobile video is widely used in the fields such as video monitoring, visual telephone and mobile TV.In order to test the effect of service control technology of the present invention, this technology is applied in the middle of certain mobile video monitor system by we.This system mainly comprises four parts: one is the front end camera device of responsible real time video collection; Two is the application servers being responsible for providing the function such as certification and management, and three is the mobile stream medium service devices being responsible for providing Video Coding and transmission; Four is the mobile terminal devices being watched real-time monitor video by mobile network.
Mobile video monitor system realizes cell phone apparatus terminal by the different local camera review of network viewing, video monitoring is not limited, as long as there is the place of cell phone network just can realize real-time video monitoring whenever and wherever possible by region and wireless network mode.In such a system, the present invention is mainly used in the fast processing of mobile terminal device to image display during video decode, for user provides mobile video service that is smooth, high-resolution.
The present invention, by mobile-terminal platform, carries out 960 × 540 high definitions amplifications to image and contrasts Riming time of algorithm.Meanwhile, on the basis of new interpolation algorithm, adopt NEON technology to carry out parallel processing optimization to it, and compare its algorithm execution efficiency.Experiment, using Wheel (480 × 270) as original input picture, amplifies twice, exports the interpolation image of 960 × 540.The time of record algorithm consumption also tabulates and compares.Error in measuring and calculating, we carry out 3 groups of experiments to often kind of algorithm respectively, and often group comprises 10000 continuous print interpolation magnification operations, and statistics often group tests the time consumed, and calculating mean consumption time and average treatment speed, concrete comparison diagram is shown in accompanying drawing 9,10.
Table 1 Wheel image is interpolation 10000 contrasts consuming time on mobile phone
Can be found out by upper table, in mobile terminal environment, new interpolation algorithm elapsed time is approximately 2 times of bicubic interpolation interpolation algorithm to 3 times, and it is per second that its interpolation 960 × 540 can reach 12 frames, can meet the requirement of mobile terminal epigraph Fast Interpolation.For the ARM framework supporting NEON technology, parallel accelerate can also be done by its NEON version to algorithm.New interpolation (Neon_New) algorithm after adopting NEON technology to accelerate, can reach the travelling speed matched in excellence or beauty with bicubic interpolation algorithm, it is per second that its interpolation 960 × 540 can reach 29 frames.
The examples of implementation of the above are only the preferred embodiment of the present invention, not limit practical range of the present invention with this, therefore the change that all shapes according to the present invention, principle are done, all should be encompassed in protection scope of the present invention.
Claims (4)
1. a mobile terminal rapid image interpolation method, is characterized in that, comprises the following steps:
1) rim detection is carried out to low resolution original image, obtain marginal information;
2) calculate the intensity at edge according to low-resolution image marginal information and human visual system, expand according to intensity edge;
3) according to the edge after expansion, image is divided into fringe region and non-edge, edge region adopts B-spline interpolation algorithm or SAI interpolation algorithm to carry out processing and preserving marginal information, adopts bicubic interpolation algorithm to process to non-edge;
4) according to existing marginal information, further Edge contrast is carried out to the edge of interpolation image, reduce marginal portion fog-level, the visual quality of prompting image;
5) interpolation algorithm is combined with NEON concurrent technique, obtains the high-definition picture of parallel optimization process.
2. a kind of mobile terminal rapid image interpolation method according to claim 1, it is characterized in that: in step 2) in, first edge strength analysis is carried out according to low-resolution image, in edge strength information, except recording position in the picture, edge, also analyze according to human visual system's edge intensity, the power according to edge strength carries out extended operation to it.
3. a kind of mobile terminal rapid image interpolation method according to claim 1, it is characterized in that: in step 3) in, by the pixel-map in target image on original image, according to edge strength information by image zoning, for the flat site of non-edge, bicubic interpolation algorithm is adopted to carry out interpolation processing, adopt B-spline interpolation algorithm or SAI interpolation algorithm with Protect edge information information for fringe region, merge the interpolation result in each region, after obtaining high resolving power interpolation image, again further edge is done to its edge and strengthen process, reach algorithm execution time efficiency and the qualitative balance of algorithm interpolation.
4. a kind of mobile terminal rapid image interpolation method according to claim 1, it is characterized in that: in step 5) in, to computation complexity, high and algorithm the module of parallelization can carry out inline assembler optimization, automatic vectorization is adopted to utilize compiler to be optimized to other modules or function, adopt inline assembler that convolution algorithm, look are melted computing, the optimization of these computings of vector operation employing NEON assembly instruction, thus make this partial code can obtain parallel processing by SIMD framework, boosting algorithm execution efficiency.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510460897.0A CN105023241A (en) | 2015-07-29 | 2015-07-29 | Fast image interpolation method for mobile terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510460897.0A CN105023241A (en) | 2015-07-29 | 2015-07-29 | Fast image interpolation method for mobile terminal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105023241A true CN105023241A (en) | 2015-11-04 |
Family
ID=54413186
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510460897.0A Pending CN105023241A (en) | 2015-07-29 | 2015-07-29 | Fast image interpolation method for mobile terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105023241A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106204441A (en) * | 2016-06-27 | 2016-12-07 | Tcl集团股份有限公司 | The method and device that a kind of image local amplifies |
| CN106878623A (en) * | 2017-04-20 | 2017-06-20 | 努比亚技术有限公司 | Photographic method, mobile terminal and computer-readable recording medium |
| CN107172364A (en) * | 2017-04-28 | 2017-09-15 | 努比亚技术有限公司 | A kind of image exposure compensation method, device and computer-readable recording medium |
| CN107273175A (en) * | 2016-04-06 | 2017-10-20 | 龙芯中科技术有限公司 | Program optimization method and device |
| CN107329889A (en) * | 2017-06-05 | 2017-11-07 | 芯海科技（深圳）股份有限公司 | A kind of method of C compilers automatic test |
| CN109145845A (en) * | 2018-08-30 | 2019-01-04 | 浦巧南 | House permission electronics resolving method |
| CN109615066A (en) * | 2019-01-30 | 2019-04-12 | 新疆爱华盈通信息技术有限公司 | A kind of method of cutting out of the convolutional neural networks for NEON optimization |
| CN111047129A (en) * | 2019-04-29 | 2020-04-21 | 泰州悦诚科技信息咨询中心 | Adaptive data correction platform |
| CN115278305A (en) * | 2022-05-12 | 2022-11-01 | 浙江大华技术股份有限公司 | Video processing method, video processing system, and storage medium |
| CN115578258A (en) * | 2022-09-29 | 2023-01-06 | 北京百度网讯科技有限公司 | Image processing method, device, equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100142842A1 (en) * | 2008-12-04 | 2010-06-10 | Harris Corporation | Image processing device for determining cut lines and related methods |
| CN102281439A (en) * | 2011-06-16 | 2011-12-14 | 杭州米加科技有限公司 | Streaming media video image preprocessing method |
| CN102622723A (en) * | 2011-05-25 | 2012-08-01 | 上海大学 | Image interpolation based on CUDA (compute unified device architecture) and edge detection |
| CN104504696A (en) * | 2014-12-16 | 2015-04-08 | 江南大学 | Embedded parallel optimization method for image salient region detection |
-
2015
- 2015-07-29 CN CN201510460897.0A patent/CN105023241A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100142842A1 (en) * | 2008-12-04 | 2010-06-10 | Harris Corporation | Image processing device for determining cut lines and related methods |
| CN102622723A (en) * | 2011-05-25 | 2012-08-01 | 上海大学 | Image interpolation based on CUDA (compute unified device architecture) and edge detection |
| CN102281439A (en) * | 2011-06-16 | 2011-12-14 | 杭州米加科技有限公司 | Streaming media video image preprocessing method |
| CN104504696A (en) * | 2014-12-16 | 2015-04-08 | 江南大学 | Embedded parallel optimization method for image salient region detection |
Non-Patent Citations (5)
| Title |
|---|
| WEI ZHE等: "Edge-contrast-guided image interpolation using directional variation field diffusion", 《PROCEEDINGS OF 2010 17TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING》 * |
| ZHANG XINGMING等: "Optimizing H.264 Decoding with Assembly Language on ARM Platform", 《COMPUTER SCIENCE & SERVICE SYSTEM,2012 INTERNATIONAL CONFERENCE》 * |
| ZHE WEI等: "Contrast-Guided Image Interpolation", 《IEEE TRANSACTIONS ON IMAGE PROCESSING》 * |
| 罗骁: "移动终端快速图像处理技术的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 陈新泽: "基于NEON引擎的JM解码器并行优化实现", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107273175A (en) * | 2016-04-06 | 2017-10-20 | 龙芯中科技术有限公司 | Program optimization method and device |
| CN106204441A (en) * | 2016-06-27 | 2016-12-07 | Tcl集团股份有限公司 | The method and device that a kind of image local amplifies |
| CN106878623A (en) * | 2017-04-20 | 2017-06-20 | 努比亚技术有限公司 | Photographic method, mobile terminal and computer-readable recording medium |
| CN107172364A (en) * | 2017-04-28 | 2017-09-15 | 努比亚技术有限公司 | A kind of image exposure compensation method, device and computer-readable recording medium |
| CN107329889A (en) * | 2017-06-05 | 2017-11-07 | 芯海科技（深圳）股份有限公司 | A kind of method of C compilers automatic test |
| CN109145845A (en) * | 2018-08-30 | 2019-01-04 | 浦巧南 | House permission electronics resolving method |
| CN109615066A (en) * | 2019-01-30 | 2019-04-12 | 新疆爱华盈通信息技术有限公司 | A kind of method of cutting out of the convolutional neural networks for NEON optimization |
| CN111047129A (en) * | 2019-04-29 | 2020-04-21 | 泰州悦诚科技信息咨询中心 | Adaptive data correction platform |
| CN115278305A (en) * | 2022-05-12 | 2022-11-01 | 浙江大华技术股份有限公司 | Video processing method, video processing system, and storage medium |
| CN115278305B (en) * | 2022-05-12 | 2024-05-07 | 浙江大华技术股份有限公司 | Video processing method, video processing system, and storage medium |
| CN115578258A (en) * | 2022-09-29 | 2023-01-06 | 北京百度网讯科技有限公司 | Image processing method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105023241A (en) | Fast image interpolation method for mobile terminal | |
| CN110136066B (en) | Video-oriented super-resolution method, device, equipment and storage medium | |
| CN111476737A (en) | Image processing method, intelligent device and computer readable storage medium | |
| CN110428382B (en) | Efficient video enhancement method and device for mobile terminal and storage medium | |
| US9615040B2 (en) | Determining a maximum inscribed size of a rectangle | |
| CN110996170A (en) | Video file playing method and related equipment | |
| CN111652830A (en) | Image processing method and device, computer readable medium and terminal equipment | |
| KR101214675B1 (en) | Method for processing a image in a handheld device and apparatus for the same | |
| CN111523652B (en) | Processor, data processing method thereof and image pickup device | |
| CN110765799A (en) | Client code scanning identification method, device, equipment and storage medium | |
| CN107920252B (en) | Panoramic video data processing method, device and system | |
| CN113190195A (en) | Method and system for visualizing large-amplitude remote sensing image | |
| US9811892B1 (en) | Separating sub-band image data for processing and merging with unprocessed image data | |
| CN110069641B (en) | Image processing method and device and electronic equipment | |
| Hung et al. | Image interpolation using convolutional neural networks with deep recursive residual learning | |
| US9928577B2 (en) | Image correction apparatus and image correction method | |
| US20150104112A1 (en) | Large Radius Edge-Preserving Low-Pass Filtering | |
| CN113689333A (en) | Image enhancement method and device | |
| CN110580274B (en) | GIS data rendering method | |
| JP2012247897A (en) | Image processing apparatus and method of processing image | |
| CN112991170A (en) | Method, device, terminal and storage medium for image super-resolution reconstruction | |
| CN111784607A (en) | Image tone mapping method, device, terminal equipment and storage medium | |
| KR20210062485A (en) | Electronic apparatus and control method thereof | |
| CN113096011B (en) | Image processing method and device and electronic equipment | |
| US20230060988A1 (en) | Image processing device and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20151104 |