CN113643341B - Different-scale target image registration method based on resolution self-adaptation - Google Patents
Different-scale target image registration method based on resolution self-adaptation Download PDFInfo
- Publication number
- CN113643341B CN113643341B CN202111184265.8A CN202111184265A CN113643341B CN 113643341 B CN113643341 B CN 113643341B CN 202111184265 A CN202111184265 A CN 202111184265A CN 113643341 B CN113643341 B CN 113643341B
- Authority
- CN
- China
- Prior art keywords
- image
- finite rank
- registration
- rank operator
- operator
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
- G06T7/337—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving reference images or patches
-
- 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
- G06T3/4007—Interpolation-based scaling, e.g. bilinear interpolation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Abstract
The invention relates to a different-scale target image registration method based on resolution self-adaptation, belonging to the field of image processing; extracting the pixel points of an image with the pixel point number of mn, and regarding the pixel points as mn-order finite rank operatorsA(ii) a After the image is amplified, if the number of the pixels of the original image cannot completely fill the amplified image, that is, the pixels of the amplified image need to be filled, and the image is scaled based on the lagrange interpolation method: for finite rank operatorAInterpolating by rows to obtain finite rank operatorBFor finite rank operatorBFinally obtaining a finite rank operator by column interpolationCFinite rank operator to be obtainedCFeeding back to the amplified image to obtain a Lagrange interpolation scaled image; and performing feature point registration by using an SIFT image registration method, and the like. The Lagrange interpolation method is combined with SIFT image registration and scale scaling to be used for registration scaling processing of the same automobile different-scale conditions in the video images, so that the pixel difference of image registration is obviously reduced.
Description
Technical Field
The invention belongs to the technical field of image data processing, and particularly relates to a different-scale target image registration method based on resolution self-adaptation.
Background
In real life, the screenshot of two front and rear video vehicles with different sizes is calculated mainly by searching the same characteristic points on the front and rear video vehicles, and the calculated proportion is used for amplifying or reducing the picture. The problem mainly exists that the scale calculated by the corresponding feature points is not necessarily capable of scaling the picture to a proper size, and is either generally enlarged or generally reduced.
For example: the method comprises the steps that a video file shot by an intersection monitoring probe obtains frame-by-frame automobile moving pictures, obviously, a closer frame image of the same automobile in different frames is slightly larger, a farther frame image of the same automobile in different frames is slightly smaller, and people hope to adaptively reduce a larger image in the front frame and the rear frame to the size same as that of a smaller image or adaptively enlarge the smaller image to the size same as that of the larger image. Due to the fact that the size of the same automobile is not consistent under different time frames, an image transformation algorithm needs to be established to stretch target images with different distances to proper sizes.
The existing image scaling method based on SIFT image registration comprises the following steps:
using SIFT image registration to obtain feature points of the image; obtaining the distance between every two characteristic points by using every two connecting lines of all the characteristic points, and averaging the distances to obtain an average proportion; and zooming the picture by using the proportion obtained by the method. However, the picture scaling effect of this method is not ideal, and the pixel value accuracy after picture scaling is low.
Therefore, at present, a method for registering different-scale target images based on resolution self-adaptation needs to be designed to solve the above problems.
Disclosure of Invention
The invention aims to provide a different-scale target image registration method based on resolution self-adaptation, which is used for solving the technical problems in the prior art, such as: using SIFT image registration to obtain feature points of the image; obtaining the distance between every two characteristic points by using every two connecting lines of all the characteristic points, and averaging the distances to obtain an average proportion; and zooming the picture by using the proportion obtained by the method. However, the picture scaling effect of this method is not ideal, and the pixel value accuracy after picture scaling is low.
In order to achieve the purpose, the technical scheme of the invention is as follows:
a different-scale target image registration method based on resolution self-adaptation comprises the following steps:
s1, acquiring image data: acquiring two images, and if the two images are different in size, zooming in and out; wherein, the number of the pixel points isThe pixel points of the image are extracted and regarded asOrder limited rank operator(ii) a After the image is enlarged, if the number of the pixels of the original image cannot completely fill the enlarged image, that is, the pixels of the enlarged image need to be filled, and the process proceeds to step S2;
s2, image scaling based on the Lagrange interpolation method: assuming that the magnified image requiresFor each pixel point, firstly, finite rank operator is subjectedInterpolating by rows to obtain finite rank operatorThen to finite rank operatorFinally obtaining a finite rank operator by column interpolationFinite rank operator to be obtainedFeedback to amplificationObtaining a Lagrange interpolation method scaled image from the later image;
s3, SIFT registration: and (4) carrying out feature point registration on the Lagrange interpolation method scaled image in the step S2 by using an SIFT image registration method, carrying out pairwise matching calculation on the obtained feature points to calculate the proportion, then solving the average value, and then carrying out image scaling on the image based on the Lagrange interpolation method once again according to the average proportion to obtain the target scaled image with different scales.
Further, in step S2, finite rank operator is subjected toInterpolating by rows to obtain finite rank operatorThe method comprises the following specific steps:
wherein the content of the first and second substances,
further, in step S2, finite rank operator is subjected toFinally obtaining a finite rank operator by column interpolationThe method comprises the following specific steps:
Wherein the content of the first and second substances,
compared with the prior art, the invention has the beneficial effects that:
the invention has the innovation points that the registration and scaling processing of the same automobile in different scales in the video images is carried out by combining the Lagrange interpolation method with SIFT image registration and scale scaling, so that the pixel difference of image registration is obviously reduced, and the problem that a smaller or larger image cannot be scaled in place is solved.
Drawings
Fig. 1 is a schematic flow chart of an embodiment of the present application.
Fig. 2 is a schematic diagram illustrating the effect of different scaling of the front and rear frames of a gray car captured by the 9 th and 10 th frames in a video image sequence according to the embodiment of the present application.
Fig. 3 is a schematic diagram illustrating the effect of different scaling of the front and rear frames of a black car captured by the 30 th and 31 th frames in the same video image sequence according to the embodiment of the present application.
Fig. 4 is a schematic diagram illustrating the effect of different scaling of the front and rear frames of a black car captured by the 40 th and 41 th frames in the same video image sequence according to the embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present invention are clearly and completely described below with reference to fig. 1 to 4 of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example (b):
as shown in fig. 1, a method for registration of an out-of-scale target image based on resolution adaptation is therefore proposed, which includes the following steps:
s1, acquiring image data: acquiring two images, and if the two images are different in size, zooming in and out; wherein, the number of the pixel points isThe pixel points of the image are extracted and regarded asOrder limited rank operator(ii) a After the image is enlarged, if the number of the pixels of the original image cannot completely fill the enlarged image, that is, the pixels of the enlarged image need to be filled, and the process proceeds to step S2;
s2, image scaling based on the Lagrange interpolation method: assuming that the magnified image requiresFor each pixel point, firstly, finite rank operator is subjectedInterpolating by rows to obtain finite rank operatorThen to finite rank operatorFinally obtaining a finite rank operator by column interpolationFinite rank operator to be obtainedFeeding back to the amplified image to obtain a Lagrange interpolation scaled image;
s3, SIFT registration: and (4) carrying out feature point registration on the Lagrange interpolation method scaled image in the step S2 by using an SIFT image registration method, carrying out pairwise matching calculation on the obtained feature points to calculate the proportion, then solving the average value, and then carrying out image scaling on the image based on the Lagrange interpolation method once again according to the average proportion to obtain the target scaled image with different scales.
Further, in step S2, finite rank operator is subjected toInterpolating by rows to obtain finite rank operatorThe method comprises the following specific steps:
wherein the content of the first and second substances,
further, in step S2, finite rank operator is subjected toFinally obtaining a finite rank operator by column interpolationThe method comprises the following specific steps:
Wherein the content of the first and second substances,
specifically, two frames of images before and after a gray car are captured in frames 9 and 10 of a video image sequence, the car occupies 248 × 243 pixels in the farther frame, and the car occupies 397 × 469 pixels in the closer frame, and the pixel values of the two are scaled to be approximately the same.
The result is shown in fig. 2, and the image registration by the method of the scheme of the application has almost no pixel difference; the SIFT image registration is adopted to have a long pixel difference of about-118 and a wide pixel difference of about 142, and the SIFT image registration is concretely characterized in that a windshield has a long pixel difference of about 21 and a wide pixel difference of about-60, a bumper has a long pixel difference of about-51 and a wide pixel difference of about-5, a vehicle lamp has a long pixel difference of about 27 and a wide pixel difference of about 39, and a vehicle head has a long pixel difference of about-86 and a wide pixel difference of about-10.
The 30 th and 31 th frames in the same video image sequence capture two frames of images of a black car, respectively occupy 254 x 254 pixels and 466 x 590 pixels, and the pixel values are scaled to be approximately the same.
The result is shown in fig. 3, and the image registration by the method of the scheme of the application has almost no pixel difference; the SIFT image registration is adopted, and a long pixel difference of about-22 and a wide pixel difference of about 3 exist, and the method is specifically characterized in that a windshield is provided with a long pixel difference of about 161 and a wide pixel difference of about 67, a bumper is provided with a long pixel difference of about 173 and a wide pixel difference of about-8, a vehicle lamp is provided with a long pixel difference of about 61 and a wide pixel difference of about 76, and a vehicle head is provided with a long pixel difference of about 42 and a wide pixel difference of about 65.
The 40 th and 41 th frames in the same video image sequence capture two frames of images of a black car, occupying 245 x 230 pixels and 398 x 422 pixels, respectively, and have pixel values scaled to be approximately as large.
The result is shown in fig. 4, and the image registration by the method of the scheme of the application has almost no pixel difference; the SIFT image registration is adopted to have a long pixel difference of about-90 and a wide pixel difference of about 20, and the SIFT image registration is concretely characterized in that a windshield has a long pixel difference of about-60 and a wide pixel difference of about 77, a bumper has a long pixel difference of about 54 and a wide pixel difference of about 54, a vehicle lamp has a long pixel difference of about 27 and a wide pixel difference of about 124, and a vehicle head has a long pixel difference of about-39 and a wide pixel difference of about 20.
Therefore, the Lagrange interpolation method is combined with SIFT image registration and scale scaling to be used for registration scaling processing of the same automobile in different scales in the video image, so that the pixel difference of image registration is remarkably reduced, and the problem that a smaller or larger image cannot be scaled in place is solved.
The above are preferred embodiments of the present invention, and all changes made according to the technical scheme of the present invention that produce functional effects do not exceed the scope of the technical scheme of the present invention belong to the protection scope of the present invention.
Claims (3)
1. A different-scale target image registration method based on resolution self-adaptation is characterized by comprising the following steps:
s1, acquiring image data: acquiring two images, and if the two images are different in size, zooming in and out; wherein, the number of the pixel points isThe pixel points of the image are extracted and regarded asOrder limited rank operator(ii) a After the image is enlarged, if the number of the pixels of the original image cannot completely fill the enlarged image, that is, the pixels of the enlarged image need to be filled, and the process proceeds to step S2;
s2, image scaling based on the Lagrange interpolation method: assuming that the magnified image requiresFor each pixel point, firstly, finite rank operator is subjectedInterpolating by rows to obtain finite rank operatorThen to finite rank operatorFinally obtaining a finite rank operator by column interpolationFinite rank operator to be obtainedFeeding back to the amplified image to obtain a Lagrange interpolation scaled image;
s3, SIFT registration: and (4) carrying out feature point registration on the Lagrange interpolation method scaled image in the step S2 by using an SIFT image registration method, carrying out pairwise matching calculation on the obtained feature points to calculate the proportion, then solving the average value, and then carrying out image scaling on the image based on the Lagrange interpolation method once again according to the average proportion to obtain the target scaled image with different scales.
2. The method for registration of different-scale target images based on resolution adaptation as claimed in claim 1, wherein in step S2, finite rank operator is appliedInterpolating by rows to obtain finite rank operatorThe method comprises the following specific steps:
wherein the content of the first and second substances,
3. the method for registration of different-scale target images based on resolution adaptation as claimed in claim 2, wherein in step S2, finite rank operator is appliedFinally obtaining a finite rank operator by column interpolationThe method comprises the following specific steps:
Wherein the content of the first and second substances,
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111184265.8A CN113643341B (en) | 2021-10-12 | 2021-10-12 | Different-scale target image registration method based on resolution self-adaptation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111184265.8A CN113643341B (en) | 2021-10-12 | 2021-10-12 | Different-scale target image registration method based on resolution self-adaptation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113643341A CN113643341A (en) | 2021-11-12 |
CN113643341B true CN113643341B (en) | 2021-12-28 |
Family
ID=78426349
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111184265.8A Active CN113643341B (en) | 2021-10-12 | 2021-10-12 | Different-scale target image registration method based on resolution self-adaptation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113643341B (en) |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102651127A (en) * | 2012-04-01 | 2012-08-29 | 深圳市万兴软件有限公司 | Image processing method and image processing system for super-resolution reconstruction |
KR20140089277A (en) * | 2012-12-31 | 2014-07-14 | 기초과학연구원 | Method for authenticating based on finger vein using SIFT keypoint |
CN106920214A (en) * | 2016-07-01 | 2017-07-04 | 北京航空航天大学 | Spatial target images super resolution ratio reconstruction method |
CN107451954A (en) * | 2017-05-23 | 2017-12-08 | 南京邮电大学 | Iterated pixel interpolation method based on image low-rank property |
CN107993258A (en) * | 2017-11-23 | 2018-05-04 | 浙江大华技术股份有限公司 | A kind of method for registering images and device |
CN109146940A (en) * | 2018-08-10 | 2019-01-04 | 山西财经大学 | A kind of high precision image method for registering based on the fitting of weight phase correlation peak |
CN110390640A (en) * | 2019-07-29 | 2019-10-29 | 齐鲁工业大学 | Graph cut image split-joint method, system, equipment and medium based on template |
CN110827332A (en) * | 2019-10-09 | 2020-02-21 | 哈尔滨工程大学 | Registration method of SAR image based on convolutional neural network |
CN112686935A (en) * | 2021-01-12 | 2021-04-20 | 武汉大学 | Airborne depth sounding radar and multispectral satellite image registration method based on feature fusion |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102479380B (en) * | 2010-11-25 | 2014-01-15 | 北京大学 | Image interpolation method capable of improving image resolution and device utilizing image interpolation method |
KR101904203B1 (en) * | 2012-06-20 | 2018-10-05 | 삼성전자주식회사 | Apparatus and method of extracting feature information of large source image using scalar invariant feature transform algorithm |
CN105809626A (en) * | 2016-03-08 | 2016-07-27 | 长春理工大学 | Self-adaption light compensation video image splicing method |
CN107330860A (en) * | 2017-07-03 | 2017-11-07 | 山东财经大学 | A kind of rational interpolation Zoom method based on CT image borders |
CN108416732A (en) * | 2018-02-02 | 2018-08-17 | 重庆邮电大学 | A kind of Panorama Mosaic method based on image registration and multi-resolution Fusion |
WO2019201336A1 (en) * | 2018-04-19 | 2019-10-24 | Shanghaitech University | Light field based reflection removal |
-
2021
- 2021-10-12 CN CN202111184265.8A patent/CN113643341B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102651127A (en) * | 2012-04-01 | 2012-08-29 | 深圳市万兴软件有限公司 | Image processing method and image processing system for super-resolution reconstruction |
KR20140089277A (en) * | 2012-12-31 | 2014-07-14 | 기초과학연구원 | Method for authenticating based on finger vein using SIFT keypoint |
CN106920214A (en) * | 2016-07-01 | 2017-07-04 | 北京航空航天大学 | Spatial target images super resolution ratio reconstruction method |
CN107451954A (en) * | 2017-05-23 | 2017-12-08 | 南京邮电大学 | Iterated pixel interpolation method based on image low-rank property |
CN107993258A (en) * | 2017-11-23 | 2018-05-04 | 浙江大华技术股份有限公司 | A kind of method for registering images and device |
CN109146940A (en) * | 2018-08-10 | 2019-01-04 | 山西财经大学 | A kind of high precision image method for registering based on the fitting of weight phase correlation peak |
CN110390640A (en) * | 2019-07-29 | 2019-10-29 | 齐鲁工业大学 | Graph cut image split-joint method, system, equipment and medium based on template |
CN110827332A (en) * | 2019-10-09 | 2020-02-21 | 哈尔滨工程大学 | Registration method of SAR image based on convolutional neural network |
CN112686935A (en) * | 2021-01-12 | 2021-04-20 | 武汉大学 | Airborne depth sounding radar and multispectral satellite image registration method based on feature fusion |
Non-Patent Citations (4)
Title |
---|
Deformable Medical Image Registration: A Survey;Aristeidis Sotiras等;《IEEE TRANSACTIONS ON MEDICAL IMAGING》;20130731;第32卷(第7期);第1153-1190页 * |
Face Verification using Scale Invariant Feature Transform with Template Security;Shubhangi Sapkal等;《2017 IEEE International Conference on Fuzzy Systems(FUZZ-IEEE)》;20170824;第1-5页 * |
低秩矩阵恢复在医学图像配准中的应用;崔宁波;《中国优秀硕士学位论文全文数据库 信息科技辑》;20151015(第10期);第I138-510页 * |
权重归一化拉格朗日插值及其空间降尺度应用;胡沛然和陈少辉;《遥感信息》;20191220;第34卷(第6期);第63-71页 * |
Also Published As
Publication number | Publication date |
---|---|
CN113643341A (en) | 2021-11-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20210174471A1 (en) | Image Stitching Method, Electronic Apparatus, and Storage Medium | |
EP3816928A1 (en) | Image super-resolution reconstruction method, image super-resolution reconstruction apparatus, and computer-readable storage medium | |
CN108765343B (en) | Image processing method, device, terminal and computer readable storage medium | |
US8000559B2 (en) | Method of correcting image distortion and apparatus for processing image using the method | |
US9308862B2 (en) | Drive assistance device | |
CN103914810B (en) | Image super-resolution for dynamic rearview mirror | |
US8570443B2 (en) | Image processing circuit and method thereof | |
US11538175B2 (en) | Method and apparatus for detecting subject, electronic device, and computer readable storage medium | |
TWI599989B (en) | Image processing method and image system for transportation | |
US20070183677A1 (en) | Dynamic range compression of high dynamic range imagery | |
JP5706594B2 (en) | Image processing device | |
US8711231B2 (en) | Digital image processing device and processing method thereof | |
US9269130B2 (en) | Image correction method using approximately non-linear regression approach and related image correction circuit | |
CN113643341B (en) | Different-scale target image registration method based on resolution self-adaptation | |
CN103632131B (en) | Apparatus and method for extracting object | |
CN106558021B (en) | Video enhancement method based on super-resolution technology | |
CN112927142B (en) | High-speed high-resolution video generation method and device based on time domain interpolation | |
Lu et al. | Two dimensional autoregressive modeling-based interpolation algorithms for image super-resolution: A comparison study | |
CN112053291B (en) | Deep learning-based low-light video noise reduction method and device | |
CN111062975B (en) | Method for accelerating real-time target detection of video frame based on perceptual hash algorithm | |
KR101969235B1 (en) | Rear vehicle sensing method by fish-eye lens and image information improvement in hard environment | |
CN110717859B (en) | Super-resolution reconstruction method based on two-way video | |
Aktar et al. | A Method for Online Interpolation of Packet-Loss Blocks in Streaming Video | |
Han et al. | Hardware and Software Co-Design Approach for Background Subtraction in High-Resolution Videos | |
CN117934408A (en) | Light field image full-reference quality evaluation method based on lens diagram |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |