CN104902178A - Parallel processing method for imaging and analyzing original data - Google Patents
Parallel processing method for imaging and analyzing original data Download PDFInfo
- Publication number
- CN104902178A CN104902178A CN201510278383.3A CN201510278383A CN104902178A CN 104902178 A CN104902178 A CN 104902178A CN 201510278383 A CN201510278383 A CN 201510278383A CN 104902178 A CN104902178 A CN 104902178A
- Authority
- CN
- China
- Prior art keywords
- imaging
- former data
- parallel processing
- data analysis
- processing
- 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
Abstract
The invention discloses a parallel processing method for imaging and analyzing original data. The method comprises the steps as follows: S1: obtaining the original data; S2, analyzing the original data with an independent algorithm processing unit to obtain an analysis result; S3: imaging an image with an independent imaging process unit to obtain an imaging result of a digital image; S4: performing information fusion to the analysis result and the imaging result to obtain a final processing result, wherein the step S2 and the step S3 are performed at the same time. The parallel processing method for imaging and analyzing the original data of the invention avoids the information loss brought by converting the digital image in the imaging process for adapting to human sense by analyzing the original data output by an image sensor and effectively improves robustness and processing precision of an image processing algorithm. The parallel processing method for imaging and analyzing original data of the invention effectively uses the advantages of parallel computing and multi-nucleation processing design, realizes parallel processing of image analysis and imaging, gets rid of limitation that the step of analyzing the digital image must be performed after the step of imaging, thereby greatly improving real time of the image processing algorithm.
Description
Technical field
The invention belongs to technical field of image processing, particularly the former data analysis of one and imaging parallel processing framework.
Background technology
Digital Image Processing converts picture signal to digital signal, and utilize computer it to be carried out to a special kind of skill of the process such as denoising, enhancing, segmentation and recovery.This technology is the important technical of the applications such as national defense safety, intelligent transportation, health care.
Traditional Digital Image Processing adopts a kind of process framework of order, and namely on the basis obtaining digital picture, then analyze image, digital picture is then be converted to through signal by the transducing signal of photoelectric device.But the development of processor is in recent years multinucleation trend, and the lifting of monokaryon frequency is not given prominence to.Therefore, what require high image resolution and high real-time in the face of practical application climbs to a higher point day by day, and due to monokaryon process frequency limited, traditional sequential processes framework must be difficult to reply.
There is researcher for the feature of polycaryon processor at present, improve at the design aspect of parser, and achieve certain effect, as " a kind of efficient parallel PCA-SIFT algorithm of multi-core processor oriented " that the people such as Wang Wei propose, (being published in National University of Defense technology's journal, the 4th phase in 2012).But above-mentioned improvement project is all carry out parallel optimization to the process structure of parser self, still the order maintaining traditional images treatment technology " first imaging; analyze again " in essence performs framework, and this makes it carry out deeper parallel optimization, is still difficult to break through efficiency bottle neck.
In addition, above-mentioned improvement project is all analyze digital picture.The 24 bit digital images generally used at present, that the former data exported by monolithic image transducer in imaging device obtain through separating mosaic, during this period, usually need to carry out necessary adjustment, in order to applicable propagation carries out compression coding toward contact needs to saturation, sharpness, contrast.Aforesaid operations will certainly cause many detailed information of former data to be tampered or to abandon, and causes irreversible information loss, thus the robustness of image processing algorithm and precision are declined to some extent.
Summary of the invention
For above defect and the technical need of prior art, the present invention is using the former data without imaging as starting point, a kind of former data analysis and imaging method for parallel processing are invented, the time consuming procedures found that brings of framework is performed with the order solving conventional digital image treatment technology, and the loss problem of the robustness about algorithm adopting digital picture to bring as analytic target and precision property.
On former data characterization imageing sensor, the number of each photosensitive unit Accumulating charge, reflects the intensity being irradiated to photosensitive unit glazed thread.Compared to digital picture, former data, without the adjustment of imaging chain to saturation, sharpness, contrast, also not through image compression encoding, remain the real information of scene to greatest extent.Therefore, the present invention utilizes the good characteristic of former data, directly using former data as analytic target, image analysis process is separated with imaging process, the parallel processing of both realizations.Its step comprises:
S1: obtain former data;
S2: carry out former data analysis by independently algorithm process unit, obtain analysis result;
S3: carry out image imaging by independently image forming process unit, obtain digital picture imaging results;
S4: described analysis result and imaging results are carried out information fusion, obtains final result.
Further, obtain former data in described step S1 to be specially: refer to that the output file by imaging device obtains the initial data of imageing sensor.
Further, the described former data that former data analysis described in described step S2 is specially obtaining carry out feature extraction and analysis, and the application combining reality carries out mating and identifying, described former data analysis comprises data processing and Data Matching, and the module of Data Matching should replace to corresponding processing method according to the applicable cases of reality.
Further, image imaging described in described step S3, refers to and carries out separating the process that mosaic processing obtains digital picture to obtained former data.
Further, described step S2 and described step S3 carries out simultaneously, i.e. described former data analysis and described image imaging parallel processing.
Further, information fusion described in described step S4 expresses in the digital picture produced and explains the result of former data analysis.
Beneficial effect of the present invention is, a kind of former data analysis provided by the invention and imaging method for parallel processing are directly for the former data analysis that imageing sensor exports, avoid in digital picture imaging process and carry out changing caused information loss for adapting to human sensory, effectively improve robustness and the processing accuracy of image processing algorithm.Simultaneously, the present invention effectively utilizes the advantage of multinucleation processor, achieves the parallel processing of graphical analysis and imaging, makes the analysis of digital picture without the need to carrying out after imaging chain, save the time understanding mosaic, thus greatly improve the real-time of image processing algorithm.
Accompanying drawing explanation
Figure 1 shows that the flow chart of a kind of former data analysis of the present invention and imaging method for parallel processing.
Embodiment
Hereafter will describe the specific embodiment of the invention in detail in conjunction with concrete accompanying drawing.It should be noted that the combination of technical characteristic or the technical characteristic described in following embodiment should not be considered to isolated, they can mutually be combined thus be reached better technique effect.
As shown in Figure 1, the invention provides a kind of former data analysis and imaging method for parallel processing, step comprises:
S1: obtain former data;
S2: carry out former data analysis by independently algorithm process unit, obtain analysis result;
S3: carry out image imaging by independently image forming process unit, obtain digital picture imaging results;
S4: described analysis result and imaging results are carried out information fusion, obtains final result.
Further, obtain former data in described step S1 to be specially: refer to the output file acquisition imageing sensor initial data by imaging device.On former data characterization imageing sensor, the number of each photosensitive unit Accumulating charge, reflects the intensity being irradiated to photosensitive unit glazed thread.Compared to digital picture, former data are without the adjustment of imaging chain to saturation, sharpness, contrast, and also not through image compression encoding, it remains the real information of scene to greatest extent.Therefore, the inventive method, directly to former data analysis, not only can shorten and separate the time overhead of mosaic process, and to avoid the people of raw information as distorting, and contributes to the precision improving process.
As shown in Figure 1, after former data acquisition, be responsible for former data analysis by route 1, route 2 is responsible for image imaging, and route 1 and route 2 parallel practice.
Further, the former data that the data analysis of step S2 Central Plains is specially obtaining carry out feature extraction and analysis, and the application combining reality carries out mating and identifying, described former data analysis comprises data processing and Data Matching, and the module of Data Matching should replace to corresponding processing method according to the applicable cases of reality.
With regard to data structure, former data are a kind of two-dimensional matrix that have recorded the intensity being irradiated to photosensitive unit glazed thread in fact.The former data be most widely used are the former data of Bayer form, i.e. Bayer Raw Data.Each location of pixels of the former data of the type only have recorded a color component in redness, green and blueness, and color component is horizontal, vertical be staggered, if odd number behavior G and B is staggered, even number line is then staggered for R and G.Because human eye is the most responsive to green, therefore in the former data of Bayer, 1/2, R and B component that G component accounts for total pixel respectively account for 1/4.
In order to make full use of colouring information abundant in former data, and the further analyzing and processing that its adaptive later stage is taked to reach the objects such as coupling, identification, therefore at former data analysis phase, first in conjunction with the special data structure of former data, data processing should be carried out to former data.Data processing in sum, is namely the suitable process that the color difference components of former data exported for imageing sensor carries out, and further, is the effective utilization to colouring information abundant in former data.
Data Matching is to reach the object of images match and the operation to feature detection, feature interpretation and characteristic matching that former data are carried out.It should be noted that, the Data Matching module in Fig. 1 has replaceability, according to the applicable cases of reality, this module should be replaced to corresponding processing method.
Image imaging described in described step S3, refers to and carries out separating the process that mosaic processing obtains digital picture to obtained former data.The general principle of separating mosaic processing is according to known color value, is recovered by the colouring information of location of pixels disappearance each in former data by the method for interpolation.Classical solution demosaicing algorithm, has bilinear interpolation, cubic spline interpolation and Kimmel interpolation method etc.On this basis, researcher improves it, and the outstanding algorithm of generation has self adaptation homogeneity to dissolve demosaicing algorithm and projections onto convex sets etc.This link can from these methods, select a kind of image quality high and the solution demosaicing algorithm that time overhead is little realizes.
Further, step S2 and step S3 carries out simultaneously, i.e. former data analysis and image imaging parallel processing.
Information fusion described in described step S4 expresses in the digital picture produced and explains the result of former data analysis.After the work of the image imaging that the former data analysis be responsible for when route 1 is responsible with route 2 completes, in the end the stage will carry out information fusion to the inventive method, namely express in the digital picture produced and explain the result of former data analysis, to reflect the effect of image procossing more intuitively.
A kind of former data analysis provided by the invention and imaging method for parallel processing are directly for the former data analysis that imageing sensor exports, avoid in digital picture imaging process and carry out changing caused information loss for adapting to human sensory, effectively improve robustness and the processing accuracy of image processing algorithm.Simultaneously, the present invention effectively utilizes the advantage of multinucleation processor, achieves the parallel processing of graphical analysis and imaging, makes the analysis of digital picture without the need to carrying out after imaging chain, save the time understanding mosaic, thus greatly improve the real-time of image processing algorithm.
Although give some embodiments of the present invention, it will be understood by those of skill in the art that without departing from the spirit of the invention herein, can change embodiment herein.Above-described embodiment is exemplary, should using embodiment herein as the restriction of interest field of the present invention.
Claims (6)
1. former data analysis and an imaging method for parallel processing, it is characterized in that, step comprises:
S1: obtain former data;
S2: carry out former data analysis by independently algorithm process unit, obtain analysis result;
S3: carry out image imaging by independently image forming process unit, obtain digital picture imaging results;
S4: described analysis result and imaging results are carried out information fusion, obtains final result.
2. a kind of former data analysis as claimed in claim 1 and imaging method for parallel processing, is characterized in that, obtains former data and be specially in described step S1: obtain imageing sensor initial data by the output file of imaging device.
3. a kind of former data analysis as claimed in claim 1 and imaging method for parallel processing, it is characterized in that, the described former data that former data analysis described in described step S2 is specially obtaining carry out feature extraction and analysis, and the application combining reality carries out mating and identifying, described former data analysis comprises data processing and Data Matching, and the module of Data Matching should replace to corresponding processing method according to the applicable cases of reality.
4. a kind of former data analysis as claimed in claim 1 and imaging method for parallel processing, is characterized in that, image imaging described in described step S3, refers to and carry out separating the process that mosaic processing obtains digital picture to obtained former data.
5. a kind of former data analysis as claimed in claim 1 and imaging method for parallel processing, is characterized in that, described step S2 and described step S3 carries out simultaneously, i.e. described former data analysis and described image imaging parallel processing.
6. a kind of former data analysis as claimed in claim 1 and imaging method for parallel processing, is characterized in that, information fusion described in described step S4, refers to and to express in the digital picture produced and to explain the result of former data analysis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510278383.3A CN104902178A (en) | 2015-05-27 | 2015-05-27 | Parallel processing method for imaging and analyzing original data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510278383.3A CN104902178A (en) | 2015-05-27 | 2015-05-27 | Parallel processing method for imaging and analyzing original data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104902178A true CN104902178A (en) | 2015-09-09 |
Family
ID=54034544
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510278383.3A Pending CN104902178A (en) | 2015-05-27 | 2015-05-27 | Parallel processing method for imaging and analyzing original data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104902178A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100104209A1 (en) * | 2008-10-25 | 2010-04-29 | Deever Aaron T | Defective color and panchromatic cfa image |
| CN102170571A (en) * | 2010-06-22 | 2011-08-31 | 上海盈方微电子有限公司 | Digital still camera framework for supporting two-channel CMOS (Complementary Metal Oxide Semiconductor) sensor |
| CN102404581A (en) * | 2011-11-02 | 2012-04-04 | 清华大学 | Color image processing method and device based on interpolation and near infrared |
| CN102665030A (en) * | 2012-05-14 | 2012-09-12 | 浙江大学 | Improved bilinear Bayer format color interpolation method |
| CN104079904A (en) * | 2014-07-17 | 2014-10-01 | 广东欧珀移动通信有限公司 | Color image generating method and device |
-
2015
- 2015-05-27 CN CN201510278383.3A patent/CN104902178A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100104209A1 (en) * | 2008-10-25 | 2010-04-29 | Deever Aaron T | Defective color and panchromatic cfa image |
| CN102170571A (en) * | 2010-06-22 | 2011-08-31 | 上海盈方微电子有限公司 | Digital still camera framework for supporting two-channel CMOS (Complementary Metal Oxide Semiconductor) sensor |
| CN102404581A (en) * | 2011-11-02 | 2012-04-04 | 清华大学 | Color image processing method and device based on interpolation and near infrared |
| CN102665030A (en) * | 2012-05-14 | 2012-09-12 | 浙江大学 | Improved bilinear Bayer format color interpolation method |
| CN104079904A (en) * | 2014-07-17 | 2014-10-01 | 广东欧珀移动通信有限公司 | Color image generating method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Griddehazenet: Attention-based multi-scale network for image dehazing | |
| Li et al. | Efficient fusion for infrared and visible images based on compressive sensing principle | |
| Timofte et al. | Ntire 2017 challenge on single image super-resolution: Methods and results | |
| TWI401963B (en) | Dynamic image compression method for face detection | |
| CN107123094B (en) | Video denoising method mixing Poisson, Gaussian and impulse noise | |
| An et al. | TR-MISR: Multiimage super-resolution based on feature fusion with transformers | |
| Ieng et al. | Asynchronous neuromorphic event-driven image filtering | |
| WO2022016350A1 (en) | Light field image processing method, light field image encoder and decoder, and storage medium | |
| US9584806B2 (en) | Using depth information to assist motion compensation-based video coding | |
| CN111800630A (en) | Method and system for reconstructing video super-resolution and electronic equipment | |
| CN103996023A (en) | Light field face recognition method based on depth belief network | |
| Niu et al. | Progressive moire removal and texture complementation for image demoireing | |
| CN104036489A (en) | Light field acquisition method | |
| CN104780383A (en) | 3D-HEVC (High Efficiency Video Coding) multi-resolution video coding method | |
| CN104715470A (en) | Klt corner point detecting device and method | |
| CN108154493B (en) | FPGA-based dual-waveband infrared image pseudo-color fusion algorithm | |
| CN103136530A (en) | Method for automatically recognizing target images in video images under complex industrial environment | |
| Sheng et al. | Guided colorization using mono-color image pairs | |
| Liang et al. | Wavelet frame based color image demosaicing | |
| CN104902178A (en) | Parallel processing method for imaging and analyzing original data | |
| CN113781339A (en) | Silicon-based multispectral signal processing method and device and mobile terminal | |
| Lu et al. | A new non-local video denoising scheme using low-rank representation and total variation regularization | |
| CN113688752B (en) | Training method, device, equipment and storage medium for face color detection model | |
| Yue et al. | Recaptured Raw Screen Image and Video Demoir\'eing via Channel and Spatial Modulations | |
| US20130163861A1 (en) | Pixel flow processing apparatus with integrated connected components labeling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150909 |
|
| RJ01 | Rejection of invention patent application after publication |