CN103258321A - Image stitching method - Google Patents
Image stitching method Download PDFInfo
- Publication number
- CN103258321A CN103258321A CN2013101769064A CN201310176906A CN103258321A CN 103258321 A CN103258321 A CN 103258321A CN 2013101769064 A CN2013101769064 A CN 2013101769064A CN 201310176906 A CN201310176906 A CN 201310176906A CN 103258321 A CN103258321 A CN 103258321A
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- line
- zone
- prime area
- 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 relates to an image stitching method. A traditional image stitching method is not good in effect. The image stitching method includes the steps: firstly, selecting a coarse registration area and carrying out pairwise weighted average on pixel values at the positions of overlapping coordinates of initial areas of two images, obtaining a reserved image, then, scanning an energy matrix of the reserved image line by line, obtaining a minimum energy line after calculation, obtaining a strip-shaped area by serving the minimum energy line to as a center to expand to two sides, respectively finding corresponding registration areas from the initial areas of the two images, carrying out image fusion on all the pixel values in the registration area of the two images to obtain a stitching area, and then finishing image stitching. According to the image stitching method, not only can an optimal matching seam be found rapidly, but also selection of an image fusion area can be reduced, and an optimized stitching image can be obtained.
Description
Technical field
The invention belongs to image data processing technology field, be specifically related to a kind of image split-joint method.
Background technology
Along with development of computer, the digitizing of various information storage is very general.In various information, the maximum of usefulness are writings and image information.Yet along with people are more and more higher to the requirement of picture quality, precision, the image splicing has become the hot issue in multimedia, Medical Image Processing and the field of Computer Graphics as an important branch of image processing techniques.Image splicing problem can be defined as the image by a series of space overlaps that align, and constitutes an image seamless, high-resolution, and it has than the higher resolution of single image and the bigger visual field.
Early stage image splicing research is used for the photograph cartography always, mainly is to taking photo by plane in a large number or the integration of the image of satellite.In recent years along with the research and development of image splicing, it becomes in conjunction with two complementary fields the drafting (IBR) based on image---the firm focus of computer vision and computer graphics, in computer vision field, the image splicing becomes the main approaches to visual scene description (Visual Scene Representations): in computing machine shape is learned, the image of real world is used for Environment in the past always, the pinup picture that namely synthesizes background and the increase synthetic body sense of reality of static state, image splicing can make IBR realistic new view of quick drafting from a series of really images.
In the night vision imaging technique of military field net, night-viewing twilight still is that infrared imaging equipment all can can't be taken wide-field picture owing to the restriction of apparatus for making a video recording, has said nothing of the annular picture of 360 degree.But in actual applications, many times need the synthetic pictures of a lot of pictures that 360 degree are captured, thereby can make the observer can observe the whole circumstances on every side.Use the image splicing, according to capture apparatus with after the situation of scenery is analyzed on every side, just can contain on every side that the multiple images of 360 degree scenery splice with what take by the shooting equipment that rotates, thereby obtain the panoramic pictures at super large visual angle or even 360 degree angles in real time.This has played very big effect in infrared early warning.
The image splicing mainly is divided into three key steps: image pre-service, image registration, image co-registration and edge smoothing.Before the image pre-service mainly refers to image registration, image being carried out pre-service such as the enhancing of squelch, texture and contrast and histogrammic normalization, is that reference diagram and search graph do not exist evident difference.Image registration mainly refers to based on image key feature or half-tone information in reference diagram and the search graph, seek best matching characteristic point or subregion, search for each subregion to or unique point between motion vector, finally estimate the overall situation between two width of cloth images, linear or nonlinear motion converter parameter.Image co-registration refers to after finishing images match, image is spliced, sews up, and smoothing processing is carried out on the border of sewing up, and allows the transition of joining edge boundary region nature.
The success or not key of image splicing is the image registration effect.Yet; usually have the different target zone and may have multiple nonlinear transformation; perhaps have large-area no obvious characteristic zone (as even grain or homochromy etc.), good image registration algorithm should be able to accurately find the corresponding position between image under multiple situation.Yet, in practical operation, can't overlap the zone on the border of two width of cloth pictures and find the desirable pixel zone that overlaps, therefore stay and may stay tangible splicing vestige or crack, have a strong impact on visual effect.This shows that image registration is core and the key of merging algorithm for images.According to the difference of method for registering images, merging algorithm for images generally can be divided into following two types:
(1). based on the relevant stitching algorithm in zone.
This is the most traditional and the most general algorithm.Be gray-scale value from image to be spliced based on the method for registering in zone, treating in the registering images zone of the same size in a zone and the reference picture uses least square method or other mathematical method to calculate the difference of its gray-scale value, this diversity ratio was judged more afterwards the similarity degree in the doubling of the image to be spliced zone, obtain scope and the position in the doubling of the image to be spliced zone thus, thereby realize the image splicing.Also can image be transformed to frequency domain by time domain by the FFT conversion, and then carry out registration.The image bigger to displacement, the mapping relations between two width of cloth images are set up in rotation that can first correcting image then.
When with the difference of two area pixel point gray-scale values during as discrimination standard, the simplest a kind of method is directly the difference of each point gray scale to be added up.This way effect is not fine, and usually variation and other reason owing to brightness, contrast causes the splicing failure.Another kind method is to calculate the related coefficient of the corresponding pixel points gray-scale value in two zones, and related coefficient is more big, and then the matching degree of two blocks of images is more high.The splicing effect of this method is much better, and success ratio increases.
(2). based on the relevant stitching algorithm of feature.
Method for registering based on feature is not the pixel value that directly utilizes image, but the feature by the pixel deduced image, be standard then with the characteristics of image, coupling is searched in the character pair zone of doubling of the image part, such stitching algorithm has than higher robustness and robustness.
Method for registering based on feature has two processes: feature extraction and feature registration.At first from two width of cloth images, extract features such as the tangible point of grey scale change, line, zone and form the feature set ridge.To exist the feature of corresponding relation to choosing as much as possible at two width of cloth image characteristic of correspondence focus utilization characteristic matching algorithms then.A series of image Segmentation Technology all is used in the extraction and Boundary Detection of feature.As canny operator, Laplce's Gauss operator, region growing.The space characteristics that extracts has closed border, opens border, cross spider and other features.The algorithm of characteristic matching has: crosscorrelation, range conversion, dynamic programming, structure matching, the relevant scheduling algorithm of chain code.
Summary of the invention
Purpose of the present invention just provides a kind of image split-joint method.
Zone to be spliced is chosen in the inventive method utilization, describes the energy matrix of scene content in the defined range image, obtains the minimum energy line, and choosing most possibly needs spliced image zone tape in the image to be spliced, finally carries out image co-registration.The such processing of image split-joint method of the present invention not only can be found Optimum Matching seam fast, and can the downscaled images integration region choose the stitching image that is optimized.
The concrete steps of the inventive method are:
Step (1). selected thick registration region: for image A to be spliced and image B, set in the image A near image B respectively
The zone as in image A prime area, the image B near image A
It is regional as the image B prime area,
=3~5;
Described image A and image B are the image of rectangle, and image A prime area and image B prime area are the zone of rectangle, and an edge of image A is an edge of image A prime area, and an edge of image B is an edge of image B prime area;
Step (2). the pixel value to each overlapping coordinate place of image A prime area and image B prime area carries out weighted mean in twos, and it is standby to obtain image C;
Step (3). the image C gray processing is handled, then the energy value of each pixel among the computed image C
Obtain the energy matrix of image C according to the energy value of each pixel in the image C;
Step (4). the energy matrix of progressive scanning picture C, with first the row all pixels as the comparison pixels point, seek the corresponding search pixel point of this comparison pixels point in eight connected regions of this comparison pixels point of next line, each search pixel point is as the comparison pixels point of next line; Described search pixel point is for being positioned at corresponding comparison pixels point next line and at the pixel of the minimum energy value of eight connected regions; The searching route of record first all pixels of row is as the candidate energies line;
Step (5). calculate the energy summation of every candidate energies line, the energy summation for the energy value of all pixels of constituting this candidate energies line and; Selecting the candidate energies line of energy summation minimum is the minimum energy line, the coordinate figure of each pixel of record minimum energy line;
Step (6). centered by the minimum energy line, the equidistant expansion to both sides respectively obtains belt-like zone, and described belt-like zone is positioned at the zone of image C; According to the coordinate of belt-like zone, in image A prime area and image B prime area, find image A registration region and image B registration region respectively, registration region is identical with the belt-like zone shape;
Corresponding pixel value in all pixel values in the image A registration region and the image B registration region is carried out image co-registration according to following formula, obtain the splicing regions of image A and image B, finish the image splicing.
In the formula
Be the final color of pixel,
,
Be respectively the color of corresponding pixel points in image A, the image B,
Be fusion coefficients,
The inventive method is different with the method for registering images of the searching unique point that generally adopts at present, overlap upward energy matrix of zone by investigating, select the minimum energy line, and centered by the minimum energy line, do equidistant expansion respectively, after obtaining belt-like zone, determine registration region, carry out the operation of image co-registration again.Adopt method of the present invention not only can find the energy line of coupling fast, and the scope of choosing that can the downscaled images integration region, raise the efficiency the stitching image that is optimized.Such processing can overlap the zone on the border of two width of cloth images find desirable coincidence pixel zone, allows the coincidence zone of two width of cloth images that the transition of smoother is arranged, and improves the visual effect of stitching image.
Embodiment
A kind of image split-joint method, concrete steps are:
Step (1). selected thick registration region: for image A to be spliced and image B, set in the image A near image B respectively
The zone as in image A prime area, the image B near the zone of image A as image B prime area ,=3~5;
Described image A and image B are the image of rectangle, and image A prime area and image B prime area are the zone of rectangle, and an edge of image A is an edge of image A prime area, and an edge of image B is an edge of image B prime area;
Step (2). the pixel value to each overlapping coordinate place of image A prime area and image B prime area carries out weighted mean in twos, and it is standby to obtain image C;
Step (3). the image C gray processing is handled, then the energy value of each pixel among the computed image C
Obtain the energy matrix of image C according to the energy value of each pixel in the image C;
Step (4). the energy matrix of progressive scanning picture C, with first the row all pixels as the comparison pixels point, seek the corresponding search pixel point of this comparison pixels point in eight connected regions of this comparison pixels point of next line, each search pixel point is as the comparison pixels point of next line; Described search pixel point is for being positioned at corresponding comparison pixels point next line and at the pixel of the minimum energy value of eight connected regions; The searching route of record first all pixels of row is as the candidate energies line;
Step (5). calculate the energy summation of every candidate energies line, the energy summation for the energy value of all pixels of constituting this candidate energies line and; Selecting the candidate energies line of energy summation minimum is the minimum energy line, the coordinate figure of each pixel of record minimum energy line;
Step (6). centered by the minimum energy line, the equidistant expansion to both sides respectively obtains belt-like zone, and described belt-like zone is positioned at the zone of image C; According to the coordinate of belt-like zone, in image A prime area and image B prime area, find image A registration region and image B registration region respectively, registration region is identical with the belt-like zone shape;
Corresponding pixel value in all pixel values in the image A registration region and the image B registration region is carried out image co-registration according to following formula, obtain the splicing regions of image A and image B, finish the image splicing.
Claims (1)
1. image split-joint method is characterized in that the concrete steps of this method are:
Step (1). selected thick registration region: for image A to be spliced and image B, set in the image A near image B respectively
The zone as in image A prime area, the image B near image A
The zone is as image B prime area ,=3~5;
Described image A and image B are the image of rectangle, and image A prime area and image B prime area are the zone of rectangle, and an edge of image A is an edge of image A prime area, and an edge of image B is an edge of image B prime area;
Step (2). the pixel value to each overlapping coordinate place of image A prime area and image B prime area carries out weighted mean in twos, and it is standby to obtain image C;
Step (3). the image C gray processing is handled, then the energy value of each pixel among the computed image C
,
Obtain the energy matrix of image C according to the energy value of each pixel in the image C;
Step (4). the energy matrix of progressive scanning picture C, with first the row all pixels as the comparison pixels point, seek the corresponding search pixel point of this comparison pixels point in eight connected regions of this comparison pixels point of next line, each search pixel point is as the comparison pixels point of next line; Described search pixel point is for being positioned at corresponding comparison pixels point next line and at the pixel of the minimum energy value of eight connected regions; The searching route of record first all pixels of row is as the candidate energies line;
Step (5). calculate the energy summation of every candidate energies line, the energy summation for the energy value of all pixels of constituting this candidate energies line and; Selecting the candidate energies line of energy summation minimum is the minimum energy line, the coordinate figure of each pixel of record minimum energy line;
Step (6). centered by the minimum energy line, the equidistant expansion to both sides respectively obtains belt-like zone, and described belt-like zone is positioned at the zone of image C; According to the coordinate of belt-like zone, in image A prime area and image B prime area, find image A registration region and image B registration region respectively, registration region is identical with the belt-like zone shape;
Corresponding pixel value in all pixel values in the image A registration region and the image B registration region is carried out image co-registration according to following formula, obtain the splicing regions of image A and image B, finish the image splicing;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2013101769064A CN103258321A (en) | 2013-05-14 | 2013-05-14 | Image stitching method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2013101769064A CN103258321A (en) | 2013-05-14 | 2013-05-14 | Image stitching method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN103258321A true CN103258321A (en) | 2013-08-21 |
Family
ID=48962215
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2013101769064A Pending CN103258321A (en) | 2013-05-14 | 2013-05-14 | Image stitching method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103258321A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104732482A (en) * | 2015-03-30 | 2015-06-24 | 中国人民解放军63655部队 | Multi-resolution image stitching method based on control points |
CN104794701A (en) * | 2014-01-21 | 2015-07-22 | 富士通株式会社 | Image splicing device, method and image processing equipment |
CN106652044A (en) * | 2016-11-02 | 2017-05-10 | 浙江中新电力发展集团有限公司 | Virtual scene modeling method and system |
CN106934762A (en) * | 2017-03-09 | 2017-07-07 | 史鹏飞 | A kind of image split-joint method and equipment |
CN106940877A (en) * | 2016-01-05 | 2017-07-11 | 富士通株式会社 | Image processing apparatus and method |
CN107146213A (en) * | 2017-05-08 | 2017-09-08 | 西安电子科技大学 | Unmanned plane image split-joint method based on suture |
CN107220955A (en) * | 2017-04-24 | 2017-09-29 | 东北大学 | A kind of brightness of image equalization methods based on overlapping region characteristic point pair |
CN108009987A (en) * | 2017-12-01 | 2018-05-08 | 中国科学院长春光学精密机械与物理研究所 | A kind of image scaling device and Zoom method |
CN108322654A (en) * | 2016-07-29 | 2018-07-24 | 广东欧珀移动通信有限公司 | Lens zoom method and apparatus and mobile terminal |
CN110097063A (en) * | 2019-04-30 | 2019-08-06 | 网易有道信息技术(北京)有限公司 | Data processing method, medium, device and the calculating equipment of electronic equipment |
CN112085650A (en) * | 2020-09-09 | 2020-12-15 | 南昌虚拟现实研究院股份有限公司 | Image processing method, image processing device, storage medium and computer equipment |
CN114359055A (en) * | 2022-03-21 | 2022-04-15 | 湖南大学 | Image splicing method and related device for multi-camera shooting screen body |
CN117333372A (en) * | 2023-11-28 | 2024-01-02 | 广东海洋大学 | Fusion splicing method of marine organism images |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1223557A2 (en) * | 2000-11-28 | 2002-07-17 | Monolith Co., Ltd. | Image interpolation method and apparatus therefor |
CN101751659A (en) * | 2009-12-24 | 2010-06-23 | 北京优纳科技有限公司 | Large-volume rapid image splicing method |
CN103020938A (en) * | 2012-12-14 | 2013-04-03 | 北京经纬恒润科技有限公司 | Method and system for stitching spatial domain images based on weighted average method |
-
2013
- 2013-05-14 CN CN2013101769064A patent/CN103258321A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1223557A2 (en) * | 2000-11-28 | 2002-07-17 | Monolith Co., Ltd. | Image interpolation method and apparatus therefor |
CN101751659A (en) * | 2009-12-24 | 2010-06-23 | 北京优纳科技有限公司 | Large-volume rapid image splicing method |
CN103020938A (en) * | 2012-12-14 | 2013-04-03 | 北京经纬恒润科技有限公司 | Method and system for stitching spatial domain images based on weighted average method |
Non-Patent Citations (1)
Title |
---|
方贤勇: "图像拼接技术研究", 《中国优秀博硕士学位论文全文数据库(博士) 信息科技辑》 * |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104794701A (en) * | 2014-01-21 | 2015-07-22 | 富士通株式会社 | Image splicing device, method and image processing equipment |
CN104732482A (en) * | 2015-03-30 | 2015-06-24 | 中国人民解放军63655部队 | Multi-resolution image stitching method based on control points |
CN104732482B (en) * | 2015-03-30 | 2018-06-12 | 中国人民解放军63655部队 | A kind of multi-resolution image joining method based on control point |
CN106940877A (en) * | 2016-01-05 | 2017-07-11 | 富士通株式会社 | Image processing apparatus and method |
CN106940877B (en) * | 2016-01-05 | 2021-04-20 | 富士通株式会社 | Image processing apparatus and method |
CN108322654B (en) * | 2016-07-29 | 2020-05-15 | Oppo广东移动通信有限公司 | Lens zooming method and device and mobile terminal |
CN108322654A (en) * | 2016-07-29 | 2018-07-24 | 广东欧珀移动通信有限公司 | Lens zoom method and apparatus and mobile terminal |
CN106652044A (en) * | 2016-11-02 | 2017-05-10 | 浙江中新电力发展集团有限公司 | Virtual scene modeling method and system |
CN106934762A (en) * | 2017-03-09 | 2017-07-07 | 史鹏飞 | A kind of image split-joint method and equipment |
CN106934762B (en) * | 2017-03-09 | 2020-08-04 | 北京星云环影科技有限责任公司 | Image splicing method and device |
CN107220955A (en) * | 2017-04-24 | 2017-09-29 | 东北大学 | A kind of brightness of image equalization methods based on overlapping region characteristic point pair |
CN107146213A (en) * | 2017-05-08 | 2017-09-08 | 西安电子科技大学 | Unmanned plane image split-joint method based on suture |
CN107146213B (en) * | 2017-05-08 | 2020-06-02 | 西安电子科技大学 | Unmanned aerial vehicle aerial image splicing method based on suture line |
CN108009987A (en) * | 2017-12-01 | 2018-05-08 | 中国科学院长春光学精密机械与物理研究所 | A kind of image scaling device and Zoom method |
CN110097063A (en) * | 2019-04-30 | 2019-08-06 | 网易有道信息技术(北京)有限公司 | Data processing method, medium, device and the calculating equipment of electronic equipment |
CN112085650A (en) * | 2020-09-09 | 2020-12-15 | 南昌虚拟现实研究院股份有限公司 | Image processing method, image processing device, storage medium and computer equipment |
CN114359055A (en) * | 2022-03-21 | 2022-04-15 | 湖南大学 | Image splicing method and related device for multi-camera shooting screen body |
CN114359055B (en) * | 2022-03-21 | 2022-05-31 | 湖南大学 | Image splicing method and related device for multi-camera shooting screen body |
CN117333372A (en) * | 2023-11-28 | 2024-01-02 | 广东海洋大学 | Fusion splicing method of marine organism images |
CN117333372B (en) * | 2023-11-28 | 2024-03-01 | 广东海洋大学 | Fusion splicing method of marine organism images |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103258321A (en) | Image stitching method | |
CN104376548B (en) | A kind of quick joining method of image based on modified SURF algorithm | |
CN101394573B (en) | Panoramagram generation method and system based on characteristic matching | |
CN101110122A (en) | Large cultural heritage picture pattern split-joint method based on characteristic | |
CN104599258A (en) | Anisotropic characteristic descriptor based image stitching method | |
CN105550981A (en) | Image registration and splicing method on the basis of Lucas-Kanade algorithm | |
Pulli et al. | Mobile panoramic imaging system | |
Ruan et al. | Image stitching algorithm based on SURF and wavelet transform | |
Shen et al. | Distortion-tolerant monocular depth estimation on omnidirectional images using dual-cubemap | |
Li et al. | Optimal seamline detection in dynamic scenes via graph cuts for image mosaicking | |
Nagaraja et al. | Parallax effect free mosaicing of underwater video sequence based on texture features | |
CN111047513A (en) | Robust image alignment method and device for cylindrical panoramic stitching | |
Shaikh et al. | Multiple feature extraction techniques in image stitching | |
Tang et al. | Sequential far infrared image mosaic using coarse-to-fine scheme | |
Jagadeeswari et al. | A comparative study based on video stitching methods | |
Sangle et al. | A method for generation of panoramic view based on images acquired by a moving camera | |
Sangle et al. | A novel approach for generation of panoramic view | |
Xia et al. | Design and implementation of tunnel image mosaic system based on open CV | |
An et al. | A Novel Fast Image Stitching Method Based on the Combination of SURF and Cell | |
Zhang et al. | Image mosaic of bionic compound eye imaging system based on image overlap rate prior | |
Lafkih et al. | image Mosaicing review: application on solar plant frames | |
Abdul-Rahim et al. | An in depth review paper on numerous image mosaicing approaches and techniques | |
Shi et al. | Image mosaics algorithm based on feature-block matching | |
CN103679722B (en) | A kind of Point matching method of normalized crosscorrelation based on irregular multiwindow | |
Huang et al. | The Research of Image Mosaic Techniques Based on Optimized SIFT Algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130821 |