CN104036477A - Large-view-field image splicing device and method based on two biomimetic eyes - Google Patents
Large-view-field image splicing device and method based on two biomimetic eyes Download PDFInfo
- Publication number
- CN104036477A CN104036477A CN201410248037.6A CN201410248037A CN104036477A CN 104036477 A CN104036477 A CN 104036477A CN 201410248037 A CN201410248037 A CN 201410248037A CN 104036477 A CN104036477 A CN 104036477A
- Authority
- CN
- China
- Prior art keywords
- image
- eyes
- definition
- devkit
- carma
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a large-view-field image splicing device and a large-view-field image splicing method based on two biomimetic eyes. A system comprises two high-definition cameras and two image fast processing modules SECO CARMA DevKit respectively connected with the two high-definition cameras, wherein the two image modules are connected onto a main control computer through switching equipment. After the high-definition cameras obtain high-definition images, the high-definition images are transmitted into the fast image processing modules SECO CARMA DevKit to be subjected to image preprocessing and feature point SIFT (Scale Invariant Feature Transform) descriptor extraction through an USB (Universal Serial Bus). The SIFT descriptor extraction mainly relies on the calculation of the maximum gradient magnitude of each CUDA (Compute Unified Device Architecture) thread and corresponding direction angles. After being extracted, the feature points of the two images are input into the main control computer through the switching equipment. The main control computer carries out feature point matching and photograph transform through a GPU (Graphics Processing Unit), and selects abutted seams for finally carrying out image splicing fusion. The application examples of the invention are mainly used for large-view-field image splicing, particularly for the image splicing technology based on the two biomimetic eyes.
Description
Technical field
The invention discloses a kind of large view field image splicing apparatus and method based on bionical eyes, belong to computer vision field, relate to coupling and the splicing of binocular image.
Background technology
At present, the image processing techniques of single camera is gradually improved, yet single camera has a lot of shortcomings, such as visual field is narrower, cannot obtain more depth information.In fields such as video monitorings, often can use a plurality of video cameras or a plurality of position angles of video camera to take to expand visual field.By the image collecting is mated and spliced, can obtain more image information.
Images match and image co-registration are two gordian techniquies of Image Mosaics.Images match is the basis of image co-registration, and the calculated amount of image matching algorithm is generally very large, so the innovation of image matching technology is depended in the development of Image Mosaics technology to a great extent.
The resolution of doing image processing on airborne platform is generally 640 * 480, for high-definition image, cannot process in real time.And because calculated amount is large, data are more, the industrial computer on bionical eyes The Cloud Terrace cannot meet the work that high-definition image is processed, mated and merges.
Summary of the invention
The object of the invention is to overcome the limitation of prior art, a kind of device and method of the large view field image splicing based on bionical eyes is provided, solve the limitation of the visual field of current bionical eyes.
In order to achieve the above object, design of the present invention is: by two high-definition cameras, gather image, the image gathering is separately imported into corresponding image fast processing module CARMA DevKit, through fast image processing, spreading out of image, by network, send in switch, finally by switch, image is sent in main control computer, in main control computer, image is carried out to images match fusion.
Large view field image splicing apparatus and method based on bionical eyes of the present invention mainly comprises:
(1) high-definition image input, two EVC-HD12U USB high-definition cameras are passed to respectively on corresponding processor by USB interface;
(2) fast image processing system: by SECO CARMA DevKit embedded type C UDA hardware and software platform parallel computing, the high-definition image of Real-time Collection is carried out to real-time image and process, extract SIFT descriptor, simultaneously by SIFT descriptor input switch.
(3) computing machine obtains real-time pictures and the SIFT descriptor gathering by switch, utilizes GPU to carry out optimum matching, projective transformation to unique point, and piece is selected and Image Mosaics merges.
According to foregoing invention design, the present invention adopts following technical proposals:
A large view field image joining method based on bionical eyes, comprises two high-definition cameras, it is characterized in that: described high-definition camera is fixedly on mobile platform, and its output is respectively connected to respectively an airborne fast processing module SECO CARMA DevKit; The output of described two airborne fast processing module SECO CARMA DevKit connects on a switch, and switch is connected to main control computer.Described two high-definition cameras are taken in after high-definition image, are made an excuse and respectively image are passed to airborne fast processing module SECO CARMA DevKit the high-definition image image of Real-time Collection is processed, and extract SIFT descriptor by USB; By switch, main control computer obtains realtime graphic and SIFT descriptor, and by describing, point mates and conversion, final realization merged the splicing of image.
A large view field image joining method based on bionical eyes, adopts and above-mentioned based on the large view field image splicing apparatus of bionical eyes, carries out Image Mosaics, it is characterized in that, splicing step is as follows:
Step is image acquisition 1.: by high-definition camera (1) (2), obtain high-definition image and import in real time image fast processing module CARMA DevKit(3 into by USB interface) (4);
Step is image and processing 2.: by CARMA DevKit(3) (4) utilize SIFT algorithm to carry out uniform feature point extraction to the roughly overlapping region of obtained image;
Step is Image Mosaics 3.: the image after fast processing is imported into switch, and import main control computer into by switch.By main control computer, utilize GPU to carry out unique point optimum matching, projective transformation, piece is selected and Image Mosaics merges.
2. the pretreated concrete steps of image are as follows for described step:
The overlapping region scope of the image to be spliced that A, basis are learnt in advance, in left image I
loverlapping region in evenly choose m*n unique point P to be matched
i(i=1,2 ..., m*n);
B, by region similarity measurement and Adaptive matching region of search in right image I
rin carry out optimum matching search, obtain optimal match point (i=1,2 ..., m*n).
Region similarity measurement is with image I to be matched
lin to be matched some P
icentered by pixel create a match window W, with the half-tone information of image in window, characterize the feature of this pixel.In image I
rin the SR of region of search, take out a neighborhood of pixels onesize with W, according to similarity measurement criterion, calculate the similarity degree between two windows.
Direction for match window is chosen, and the matching algorithm under traditional rectangular window is mainly to mate for the stereo-picture of proofreading and correct through polar curve, does not have rotational invariance.And for Image Mosaics, between image corresponding point to be spliced, may exist rotational transform, for addressing this problem, need the gradient of calculated characteristics point.
Utilize each thread block of CUDA to comprise 256 * 1 * 1 thread, gradient of each thread computes, mould and direction, information with shared drive Shared Memory storage pixel on high-speed chip, calculate like this amplitude and three deflections of the gradient of three passages, now get the channel value of gradient magnitude maximum and corresponding deflection, concrete calculating undertaken by following formula:
Wherein
,
represent respectively x
,the gradient of y direction, θ represents the principal direction of match window.
The concrete steps that 3. described step utilizes GPU to splice by main control computer in Image Mosaics are as follows:
1) select to be matched some P
i(i=1,2 ..., m*n), utilize algorithm in the SR of region of search, to search corresponding match point;
2) utilize the conversion of photographing of projective transformation model;
3) select suitable piece to carry out Image Mosaics fusion.
The present invention compared with prior art, has following apparent cynapse substantive distinguishing features and significantly technical progress:
The present invention gathers high-definition image by two image input systems, then image is passed to SECO CARMA DevKit embedded type C UDA hardware and software platform the high-definition image of Real-time Collection is carried out to image processing, extract SIFT descriptor, by switch, be input to that computing machine is described son coupling and image co-registration is spliced.Compared with prior art, operand of the present invention is less, and data volume is less, processes fast, can obtain the depth information of large visual field.
Accompanying drawing explanation
Fig. 1 the present invention is based on the large view field image splicing apparatus block diagram of bionical eyes
Fig. 2 the present invention is based on the flow chart of the large view field image joining method of bionical eyes
Embodiment
Below in conjunction with accompanying drawing, the preferred embodiment in the present invention is clearly and completely described, obviously, described example is only a part of example of the present invention.
Embodiment mono-:
Referring to Fig. 1, this large view field image splicing apparatus based on freeing eyes, comprise two high-definition cameras (1,2), it is characterized in that: described two high-definition cameras (1,2) are connected respectively to corresponding image fast processing module CARMA DevKit(3,4); Described image fast processing module CARMA DevKit(3,4) by network, be connected to switch (5); Described switch (5) is connected on main control computer (6).
Embodiment bis-:
Referring to Fig. 2, this large view field image joining method based on bionical eyes, it is as follows that employing said apparatus carries out Image Mosaics concrete operation step:
Step 1 image acquisition: obtain high-definition image by high-definition camera (1,2) and import in real time image fast processing module CARMA DevKit(3,4 into by USB interface);
Step 2 image pre-service: by CARMA DevKit(3,4) utilize SIFT algorithm to carry out uniform feature point extraction to the roughly overlapping region of obtained image, the concrete steps of SIFT descriptor feature point extraction are as follows:
The overlapping region scope of the image to be spliced that A, basis are learnt in advance, in left image I
loverlapping region in evenly choose m*n unique point P to be matched
i(i=1,2 ..., m*n);
B, by region similarity measurement and Adaptive matching region of search, in right image I R, carry out optimum matching search, obtain optimal match point (i=1,2 ..., m*n).
Region similarity measurement is with image I to be matched
lin to be matched some P
icentered by pixel create a match window W, with the half-tone information of image in window, characterize the feature of this pixel.In image I
rin the SR of region of search, take out a neighborhood of pixels onesize with W, according to similarity measurement criterion, calculate the similarity degree between two windows.
Direction for match window is chosen, and the matching algorithm under traditional rectangular window is mainly to mate for the stereo-picture of proofreading and correct through polar curve, does not have rotational invariance.And for Image Mosaics, between image corresponding point to be spliced, may exist rotational transform, for addressing this problem, need the gradient of calculated characteristics point.
Utilize each thread block of CUDA to comprise 256 * 1 * 1 thread, gradient of each thread computes, mould and direction, information with shared drive Shared Memory storage pixel on high-speed chip, calculate like this amplitude and three deflections of the gradient of three passages, now get the channel value of gradient magnitude maximum and corresponding deflection, concrete calculating undertaken by following formula:
Wherein
,
represent respectively x, the gradient of y direction,
the principal direction that represents match window.
Step 3 Image Mosaics: the image after fast processing is imported into switch, and import main control computer into by switch.By main control computer, utilize GPU to carry out unique point optimum matching, projective transformation, piece is selected and Image Mosaics merges:
1) select to be matched some P
i(i=1,2 ..., m*n), utilize algorithm in the SR of region of search, to search corresponding match point;
2) utilize the conversion of photographing of projective transformation model;
3) select suitable piece to carry out Image Mosaics fusion.
The above, be only the specific embodiment of the present invention, but protection scope of the present invention is not only confined to this, and within the scope of the method that any similar experiment discloses in the present invention, the variation that can expect easily and replacement, within all should being encompassed in protection scope of the present invention.Therefore, protection scope of the present invention should be the described protection domain with claim and is as the criterion.
Claims (4)
1. the large view field image splicing apparatus based on bionical eyes, comprise two high-definition cameras (1,2), it is characterized in that: described two high-definition cameras (1,2) are connected respectively to corresponding image fast processing module CARMA DevKit(3,4); Described image fast processing module CARMA DevKit(3,4) by network, be connected to switch (5); Described switch (5) is connected on main control computer (6).
2. the large view field image joining method based on bionical eyes, adopts and according to claim 1ly based on the large view field image splicing apparatus of bionical eyes, carries out Image Mosaics, it is characterized in that splicing step as follows:
Step is image acquisition 1.: by high-definition camera (1,2), obtain high-definition image and import in real time image fast processing module CARMA DevKit(3,4 into by USB interface);
2. image pre-service of step: by CARMA DevKit(3,4) utilize SIFT algorithm to carry out uniform feature point extraction to the roughly overlapping region of obtained image;
Step is Image Mosaics 3.: by main control computer, utilize GPU to carry out unique point optimum matching, projective transformation, piece is selected and Image Mosaics merges.
3. the large view field image joining method based on bionical eyes according to claim 2, is characterized in that, described step 2. image pre-service concrete steps is as follows:
The overlapping region scope of the image to be spliced that A, basis are learnt in advance is evenly chosen m*n unique point Pi to be matched in the overlapping region of left image I L, i=1, and 2 ..., m*n;
B, by region similarity measurement and Adaptive matching region of search, in right image I R, carry out optimum matching search, obtain optimal match point, i=1,2 ..., m*n;
Region similarity measurement be in image I L to be matched centered by be matched some Pi pixel create a match window W, with the half-tone information of image in window, characterize the feature of this pixel; In image I R region of search SR, take out a neighborhood of pixels onesize with W, according to similarity measurement criterion, calculate the similarity degree between two windows;
Direction for match window is chosen, and the matching algorithm under traditional rectangular window is mainly to mate for the stereo-picture of proofreading and correct through polar curve, does not have rotational invariance; And for Image Mosaics, between image corresponding point to be spliced, may exist rotational transform, for addressing this problem, need the gradient of calculated characteristics point;
Utilize each thread block of CUDA to comprise 256 * 1 * 1 thread, gradient of each thread computes, mould and direction, information with shared drive Shared Memory storage pixel on high-speed chip, calculate like this amplitude and three deflections of the gradient of three passages, now get the channel value of gradient magnitude maximum and corresponding deflection, concrete calculating undertaken by following formula:
Wherein
,
represent respectively
x, ythe gradient of direction,
the principal direction that represents match window.
4. the large view field image joining method based on bionical eyes according to claim 2, is characterized in that the concrete steps that 3. described step utilize GPU to splice by main control computer (6) in Image Mosaics are as follows:
1) select to be matched some P
i(i=1,2 ..., m*n), utilize algorithm in the SR of region of search, to search corresponding match point;
2) utilize the conversion of photographing of projective transformation model;
3) select suitable piece to carry out Image Mosaics fusion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410248037.6A CN104036477A (en) | 2014-06-06 | 2014-06-06 | Large-view-field image splicing device and method based on two biomimetic eyes |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410248037.6A CN104036477A (en) | 2014-06-06 | 2014-06-06 | Large-view-field image splicing device and method based on two biomimetic eyes |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104036477A true CN104036477A (en) | 2014-09-10 |
Family
ID=51467237
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410248037.6A Pending CN104036477A (en) | 2014-06-06 | 2014-06-06 | Large-view-field image splicing device and method based on two biomimetic eyes |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104036477A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104599317A (en) * | 2014-12-18 | 2015-05-06 | 深圳市亿思达科技集团有限公司 | Mobile terminal and method for achieving 3D (three-dimensional) scanning modeling function |
| CN109255754A (en) * | 2018-09-30 | 2019-01-22 | 北京宇航时代科技发展有限公司 | A kind of large scene polyphaser image mosaic and the method and system really showed |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201488954U (en) * | 2009-08-25 | 2010-05-26 | 周良川 | Plate surface quality detection system |
| CN202512133U (en) * | 2012-03-29 | 2012-10-31 | 河海大学 | Large-scale particle image velocity measuring system based on double-camera visual field splice |
| CN103179401A (en) * | 2013-03-19 | 2013-06-26 | 燕山大学 | Processing method and device for multi-agent cooperative video capturing and image stitching |
| CN103473532A (en) * | 2013-09-06 | 2013-12-25 | 上海大学 | Pedestrian fast detection system and method of vehicle-mounted platform |
| US20140064626A1 (en) * | 2012-08-30 | 2014-03-06 | Samsung Techwin Co., Ltd. | Adaptive image processing apparatus and method based in image pyramid |
-
2014
- 2014-06-06 CN CN201410248037.6A patent/CN104036477A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201488954U (en) * | 2009-08-25 | 2010-05-26 | 周良川 | Plate surface quality detection system |
| CN202512133U (en) * | 2012-03-29 | 2012-10-31 | 河海大学 | Large-scale particle image velocity measuring system based on double-camera visual field splice |
| US20140064626A1 (en) * | 2012-08-30 | 2014-03-06 | Samsung Techwin Co., Ltd. | Adaptive image processing apparatus and method based in image pyramid |
| CN103179401A (en) * | 2013-03-19 | 2013-06-26 | 燕山大学 | Processing method and device for multi-agent cooperative video capturing and image stitching |
| CN103473532A (en) * | 2013-09-06 | 2013-12-25 | 上海大学 | Pedestrian fast detection system and method of vehicle-mounted platform |
Non-Patent Citations (1)
| Title |
|---|
| 郭一汉: "基于CUDA的实时图像拼接技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104599317A (en) * | 2014-12-18 | 2015-05-06 | 深圳市亿思达科技集团有限公司 | Mobile terminal and method for achieving 3D (three-dimensional) scanning modeling function |
| CN104599317B (en) * | 2014-12-18 | 2017-10-31 | 深圳市魔眼科技有限公司 | A kind of mobile terminal and method for realizing 3D scanning modeling functions |
| CN109255754A (en) * | 2018-09-30 | 2019-01-22 | 北京宇航时代科技发展有限公司 | A kind of large scene polyphaser image mosaic and the method and system really showed |
| CN109255754B (en) * | 2018-09-30 | 2023-03-14 | 北京宇航时代科技发展有限公司 | Method and system for splicing and really displaying large-scene multi-camera images |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11468585B2 (en) | Pseudo RGB-D for self-improving monocular slam and depth prediction | |
| Zhan et al. | Unsupervised learning of monocular depth estimation and visual odometry with deep feature reconstruction | |
| CN110096925B (en) | Enhancement method, acquisition method and device of facial expression image | |
| CN101859433B (en) | Image mosaic device and method | |
| Tanskanen et al. | Live metric 3D reconstruction on mobile phones | |
| CN103198477B (en) | Apple fruitlet bagging robot visual positioning method | |
| CN111151463B (en) | Mechanical arm sorting and grabbing system and method based on 3D vision | |
| US8179448B2 (en) | Auto depth field capturing system and method thereof | |
| CN101145201A (en) | Quick target identification and positioning system and method | |
| CN108073857A (en) | The method and device of dynamic visual sensor DVS event handlings | |
| CN103413139A (en) | Electric equipment abnormal heating detection method based on infrared inspection video data of power line inspection | |
| CN105282513A (en) | Device and method for detecting operation state of ultra-high-voltage transformer in transformer substation based on 3D infrared panoramic image | |
| CN104677330A (en) | Small binocular stereoscopic vision ranging system | |
| CN103942843A (en) | Fairway and ship three-dimensional model dynamic presenting method based on video | |
| Nocerino et al. | 3D reconstruction with a collaborative approach based on smartphones and a cloud-based server | |
| CN111127522B (en) | Depth optical flow prediction method, device, equipment and medium based on monocular camera | |
| CN113015978A (en) | Processing images to locate novel objects | |
| CN111179330A (en) | Binocular vision scene depth estimation method based on convolutional neural network | |
| CN104457758A (en) | Video-acquisition-based Visual Map database establishing method and indoor visual positioning method using database | |
| TW201826223A (en) | Depth map generation device | |
| CN111047636B (en) | Obstacle avoidance system and obstacle avoidance method based on active infrared binocular vision | |
| CN101370088A (en) | Scene matching apparatus and method for virtual studio | |
| CN103707300A (en) | Manipulator device | |
| CN115008454A (en) | Robot online hand-eye calibration method based on multi-frame pseudo label data enhancement | |
| CN104036477A (en) | Large-view-field image splicing device and method based on two biomimetic eyes |
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 |
Application publication date: 20140910 |
|
| WD01 | Invention patent application deemed withdrawn after publication |