CN105825476B - A Fast Image Scanning and Splicing Method Applied in DSP - Google Patents
A Fast Image Scanning and Splicing Method Applied in DSP Download PDFInfo
- Publication number
- CN105825476B CN105825476B CN201610409070.1A CN201610409070A CN105825476B CN 105825476 B CN105825476 B CN 105825476B CN 201610409070 A CN201610409070 A CN 201610409070A CN 105825476 B CN105825476 B CN 105825476B
- Authority
- CN
- China
- Prior art keywords
- image
- line
- reference line
- pixel
- matched
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 11
- 238000001514 detection method Methods 0.000 description 2
- 238000006073 displacement reaction Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- 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
- G06T2207/10004—Still image; Photographic image
- G06T2207/10008—Still image; Photographic image from scanner, fax or copier
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
本发明公开一种应用于DSP的图像快速扫描拼接方式，其包括如下步骤：步骤1：获取待拼接的图像集合，选取第一张图像和第二张图像进行拼接：首先确定基准线和匹配范围；步骤2：计算基准线上相邻像素之差；步骤3：在匹配线上确定最佳匹配位置，其中匹配线处于匹配范围内，且与基准线具有相同的高度；步骤4：实现拼接：找到最佳匹配位置之后将第二张图像在匹配线右侧的部分替换到第一张图像基准线的右侧，即得到拼接结果图像；步骤5：接下来将步骤4得到的拼接结果图像作为第一张图像，接着读取一张新图像作为第二张图像重复步骤1‑步骤4，即可实现对多张图像进行拼接。The invention discloses an image fast scanning and splicing method applied to DSP, which comprises the following steps: Step 1: acquiring a set of images to be spliced, selecting a first image and a second image for splicing: first determining a reference line and a matching range ; Step 2: Calculate the difference between adjacent pixels on the reference line; Step 3: Determine the best matching position on the matching line, where the matching line is within the matching range and has the same height as the reference line; Step 4: Implement stitching: After finding the best matching position, replace the part of the second image on the right side of the matching line to the right side of the reference line of the first image to obtain the stitching result image; Step 5: Next, use the stitching result image obtained in Step 4 as The first image, then read a new image as the second image and repeat steps 1-4 to achieve stitching of multiple images.
Description
Technical field
The present invention relates to technical field of image processing, and in particular to a kind of quick Overlap-scanning side of image applied to DSP
Formula.
Background technique
Currently, image mosaic is typically necessary carry out images match.The matching process of image generally has matching based on gray scale
Standard, registration based on feature etc..Existing frequently-used mode is all using the registration mode based on feature.It is existing to be matched based on feature
In the technical solution of quasi- mode, figure generally is realized using Harris Corner Detection, SIFT scale invariant feature conversion scheduling algorithm
As splicing.Wherein Harris Corner Detection, principle are to detect characteristic point using the feature of a second-order matrix, are extracted
Feature have rotational invariance, the disadvantage is that not having scale invariability；SIFT algorithm is to detect certain by Gauss difference space
Check that certain extreme values carry out location feature point.SIFT algorithm improves the quality and accuracy of images match, using also relatively wider
It is general.
However the existing preferable image mosaic mode of imaging effect needs to realize in high-performance PC machine, needs efficiently to locate
Manage speed and very big memory headroom, but the embedded processing systems of similar dsp processor, suffer from processing speed and
The limitation of memory headroom, it is undesirable in terms of processing speed and splicing effect using conventional image mosaic mode.
Summary of the invention
Therefore, for above-mentioned problem, the present invention is based on the above-mentioned requirement to dsp system, propose that one kind quickly scans
Panoramic mosaic mode improves existing joining method, reduces the limitation of processing speed and memory headroom.
In order to solve the above-mentioned technical problem, thinking of the present invention is as follows: when the distance moved between two pictures
It when excessive, then can cannot be neglected deformation between image, may can also be especially for user with the variation of light and shade
Center rotates to the right the case where shooting panoramic photo.At this time to obtain ideal splicing result, two pictures are not only found out
Match point will also carry out deformation correction and Lightness disposal to picture.However, it is contemplated that the moving distance between two pictures is minimum
When (for example only moving a column pixel), then the shooting interval of two pictures is extremely short, and the deformation and light and shade between image are poor
It is different all negligible, so that it may the image mosaic mode quickly scanned using one kind.
Specifically, the technical scheme adopted by the invention is that, a kind of quick Overlap-scanning mode of image applied to DSP,
Include the following steps:
Step 1: obtaining image collection to be spliced, choose first image and second image is spliced: first really
Determine reference line and matching range: choosing first image and second image in image collection, first looked in first image
One reference line, up-and-down boundary of the reference line from image respectively vacate 10 row pixels, high 460 pixel, and the right margin 320 from image arranges
Pixel；Then it keeps left second image center and determines a matching range, matching range size is wide 30 column pixel, high by 470
Pixel, up-and-down boundary of the matching range from image respectively vacate 5 row pixels；
Step 2: the difference of adjacent pixel on calculating benchmark line: in matching range by matched line one by one with reference line carry out
Match, wherein matched line and reference line have same height, and matching process is: two neighboring pixel value is first found out in reference line
Maximum preceding 50 pairs of pixels, and record the position where this 50 pairs of values:
A1=a1-a2；
A2=a2-a3；
……
A459=a459-a460；
The wherein pixel value that a [i] is every on benchmark line, A [i] is the difference of its adjacent two o'clock, 1≤i≤459.To A
[i] carries out sequence from big to small and records preceding 50 value A [0]~corresponding coordinate position of A [49]；
Step 3: best match position is determined on a match line: and then found out two-by-two on corresponding position in matched line
The difference of pixel value, and 50 pairs of differences of this 50 pairs of differences and reference line are subtracted each other two-by-two and the sum of are taken absolute value, to matching model
Each matched line in enclosing repeats above operation, then the smallest matched line of the sum of absolute value is the best match position to be looked for
It sets.Ask:
Wherein AiFor the pixel value in benchmark line, BiFor the pixel value in matched line.It finds out corresponding to minimum value
I position, as best match position；
Step 4: realizing splicing: finding best match position and be later substituted into second image in matched line right part
To get splicing result image on the right side of first image benchmark line；
Step 5: the splicing result image for next obtaining step 4 then reads a new figure as first image
Multiple images can be spliced as continuing above step as second image to obtain panorama sketch；
Step 6: cut up-and-down boundary: the panoramic picture obtained at this time has good effect, however due to flat in camera
Upper lower deviation is inevitable during moving, therefore result images can be needed there are blank by upper and lower gutter on up-and-down boundary
Divide and cuts out to get final result image.
Panoramic mosaic is carried out to image using the above method of the invention, speed is fast, slowly revolves in user's handheld device
When turning, a new images can be read every a few tens of milliseconds and be spliced, therefore the difference between the image of adjacent twi-read
Very little is just as above-mentioned example, so the effect and reality degree of splicing are also high.To sum up, the present invention is using above-mentioned
Method enormously simplifies Processing Algorithm, to greatly reduce the limit of the processing speed and memory headroom to dsp processor
System, disclosure is particularly well suited to hand held devices to carry out image mosaic using above-mentioned quick scanning mode, has expanded it significantly
Application scenarios have good practicability.
Detailed description of the invention
Now in conjunction with specific embodiment, the present invention is further described.
During image mosaic, when the distance moved between two pictures is excessive, then it can not allow to neglect between image
Depending on deformation, may can also with light and shade variation, the feelings of shooting panoramic photo are rotated to the right especially for customer-centric
Condition.At this time to obtain ideal splicing result, the match point of two pictures is not only found out, also deformation is carried out to picture and rectified
Just and Lightness disposal.However, it is contemplated that when the moving distance between two pictures is minimum (for example only moving a column pixel), then
The shooting interval of two pictures is extremely short, all negligible (picture of this project of the deformation and light and shade difference between image
Size is fixed as 640*480), such as two picture moving distances only have 10 column pixels, so that it may it is quickly scanned using one kind
Image mosaic mode.Connecting method of the invention is applied to using DSP as the embedded device of hardware carrier, which can
It to carry out the Image Acquisition of 30 frame per second, therefore is actually carrying out in splicing, in pixel per second movement in 300 column left and right
In the case of the image mosaic set forth in the present invention quickly scanned can be realized.
Specifically, a kind of quick Overlap-scanning mode of image applied to DSP of the invention, the specific implementation steps are as follows:
Step 1: obtaining image collection to be spliced first, then determine reference line matching range: choosing in image collection
Two images, first a reference line is looked in first image, up-and-down boundary of the reference line from image respectively vacates 10 row pixels,
High 460 pixel, why vacating when 10 row pixels allow for the splicing of adjacent two picture has the case where displacement up and down, from image
320 column pixel of right margin, why image middle position look for reference line be because shooting when image hit exactly caused by shape
Become minimum；Then it keeps left second image center and determines a matching range, range size is wide 30 column pixel, high 470 picture
Element, matching range respectively vacate 5 row pixels from image up-and-down boundary, so just can be corrected the case where upper and lower displacement to realize
Splicing；
Step 2: the difference of adjacent pixel on calculating benchmark line: the core concept of algorithm be in matching range by matched line by
Item is matched with reference line, and wherein matched line and reference line have same height, and matching process is: first asking in reference line
The maximum preceding 50 pairs of pixels of two neighboring pixel value out, and record the position where this 50 pairs of values:
A1=a1-a2；
A2=a2-a3；
……
A459=a459-a460；
The wherein pixel value that a [i] is every on benchmark line, A [i] is the difference of its adjacent two o'clock, 1≤i≤459.To A
[i] carries out sequence from big to small and records preceding 50 value A [0]~corresponding coordinate position of A [49]；
Step 3: determining best match position on a match line: finding out pixel two-by-two on corresponding position in matched line
The difference of value, and 50 pairs of differences of this 50 pairs of differences and reference line are subtracted each other two-by-two and the sum of are taken absolute value, in matching range
Each matched line repeat above operation, then the smallest matched line of the sum of absolute value be the best match position to be looked for.I.e.
It asks:
Wherein AiFor the pixel value in benchmark line, BiFor the pixel value in matched line.It finds out corresponding to minimum value
I position, as best match position；
Step 4: realizing splicing: finding best match position and be later substituted into second image in matched line right part
To get splicing result image on the right side of first image benchmark line；
Step 5: next using result images as first image, then reading a new images as second image
Multiple images can be spliced to obtain panorama sketch by continuing above step；
Step 6: cut up-and-down boundary: the panoramic picture obtained at this time has good effect, however due to flat in camera
Upper lower deviation is inevitable during moving, therefore result images can be needed there are blank by upper and lower gutter on up-and-down boundary
Divide and cuts out to get final result image.
Panoramic mosaic should be carried out to image in this way, speed is fast, can be with when user's handheld device slowly rotates
Every a few tens of milliseconds read a new images spliced, therefore the difference very little between the image of adjacent twi-read just as with
On the example mentioned it is the same, so the effect and reality degree of splicing are also high.
Although specifically showing and describing the present invention in conjunction with preferred embodiment, those skilled in the art should be bright
It is white, it is not departing from the spirit and scope of the present invention defined by the appended claims, it in the form and details can be right
The present invention makes a variety of changes, and is protection scope of the present invention.
Claims (3)
1. a kind of quick Overlap-scanning mode of image applied to DSP, characterized by the following steps:
Step 1: obtaining image collection to be spliced, choose first image and second image is spliced: determining base first
Directrix and matching range；
Step 2: the difference of adjacent pixel on calculating benchmark line；
Step 3: determining best match position on a match line, wherein matched line is in matching range, and is had with reference line
Identical height；
Step 4: realizing splicing: finding partial replacement after best match position by second image on the right side of matched line to the
The right side of one image benchmark line to get arrive splicing result image；
Step 5: the splicing result image for next obtaining step 4 then reads a new images and makees as first image
Step 1- step 4 is repeated for second image, can be realized and multiple images are spliced；
The step 3 specifically includes following content: matched line matched with reference line one by one in matching range, wherein
Wiring and reference line have same height, and matching process is: first found out in reference line two neighboring pixel value it is maximum before
50 pairs of pixels, and record the position where this 50 pairs of values:
A1=a1-a2；
A2=a2-a3；
……
A459=a459-a460；
The wherein pixel value that a [i] is every on benchmark line, A [i] is the difference of its adjacent two o'clock, 1≤i≤459；
Sequence from big to small is carried out to A [i], and records preceding 50 value A [0]~corresponding coordinate position of A [49]；
Then the difference of pixel value two-by-two is found out on corresponding position in matched line, and by the 50 of this 50 pairs of differences and reference line
Difference is subtracted each other two-by-two and the sum of is taken absolute value, each matched line in matching range is repeated above operation, then absolutely
The smallest matched line of the sum of value is the best match position to be looked for；Ask:
Wherein AiFor the pixel value in benchmark line, BiFor the pixel value in matched line；Find out i corresponding to minimum value
Position, as best match position.
2. the quick Overlap-scanning mode of image according to claim 1, it is characterised in that: further include step 6: to what is obtained
The up-and-down boundary of final splicing result image is cut.
3. the quick Overlap-scanning mode of image according to claim 1 or 2, it is characterised in that: the step 1 specifically includes
Following content: choosing first image and second image in image collection, first look for a reference line in first image,
Up-and-down boundary of the reference line from image respectively vacates 10 row pixels, high 460 pixel, the 320 column pixel of right margin from image；Then exist
Second image center, which keeps left, determines a matching range, and matching range size is wide 30 column pixel, and high 470 pixel matches model
It encloses the up-and-down boundary from image and respectively vacates 5 row pixels.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610409070.1A CN105825476B (en) | 2016-06-12 | 2016-06-12 | A Fast Image Scanning and Splicing Method Applied in DSP |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610409070.1A CN105825476B (en) | 2016-06-12 | 2016-06-12 | A Fast Image Scanning and Splicing Method Applied in DSP |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105825476A CN105825476A (en) | 2016-08-03 |
| CN105825476B true CN105825476B (en) | 2019-02-05 |
Family
ID=56532250
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610409070.1A Expired - Fee Related CN105825476B (en) | 2016-06-12 | 2016-06-12 | A Fast Image Scanning and Splicing Method Applied in DSP |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105825476B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110555803B (en) * | 2019-09-12 | 2023-08-08 | 福建云造科技有限公司 | Method for judging whether patterns in printing can be spliced or not and pattern splicing method |
| CN112102215A (en) * | 2020-09-03 | 2020-12-18 | 广州南沙联合集装箱码头有限公司 | Image fast splicing method based on error statistics |
| CN115578441B (en) * | 2022-08-30 | 2023-07-28 | 感知信息科技(浙江)有限责任公司 | Vehicle side image stitching and vehicle size measuring method based on deep learning |
| CN116320294A (en) * | 2023-02-15 | 2023-06-23 | 广东电网有限责任公司 | Electric power operation site panoramic security monitoring cavity compensation method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6785427B1 (en) * | 2000-09-20 | 2004-08-31 | Arcsoft, Inc. | Image matching using resolution pyramids with geometric constraints |
| CN101398901A (en) * | 2008-10-31 | 2009-04-01 | 中国航空无线电电子研究所 | Rapid image matching method for auxiliary navigation |
| CN102908122A (en) * | 2012-09-17 | 2013-02-06 | 广州市伟迈机电科技有限公司 | Imaging and image mosaic processing methods for digitalizer |
| CN103617625A (en) * | 2013-12-13 | 2014-03-05 | 中国气象局北京城市气象研究所 | Image matching method and image matching device |
-
2016
- 2016-06-12 CN CN201610409070.1A patent/CN105825476B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6785427B1 (en) * | 2000-09-20 | 2004-08-31 | Arcsoft, Inc. | Image matching using resolution pyramids with geometric constraints |
| CN101398901A (en) * | 2008-10-31 | 2009-04-01 | 中国航空无线电电子研究所 | Rapid image matching method for auxiliary navigation |
| CN102908122A (en) * | 2012-09-17 | 2013-02-06 | 广州市伟迈机电科技有限公司 | Imaging and image mosaic processing methods for digitalizer |
| CN103617625A (en) * | 2013-12-13 | 2014-03-05 | 中国气象局北京城市气象研究所 | Image matching method and image matching device |
Non-Patent Citations (1)
| Title |
|---|
| 基于DSP的实时图像拼接技术;张云峰;《液晶与显示》;20131231;第28卷(第6期);963-967 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105825476A (en) | 2016-08-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8831382B2 (en) | Method of creating a composite image | |
| CN101416219B (en) | Foreground/background segmentation in digital images | |
| CN105825476B (en) | A Fast Image Scanning and Splicing Method Applied in DSP | |
| US20100142807A1 (en) | Image identification method and imaging apparatus | |
| CN103327245B (en) | A kind of Atomatic focusing method of infrared imaging system | |
| CN112215925A (en) | Self-adaptive follow-up tracking multi-camera video splicing method for coal mining machine | |
| WO2017092631A1 (en) | Image distortion correction method for fisheye image, and calibration method for fisheye camera | |
| CN110136083B (en) | A method and device for updating basemaps combined with interactive | |
| CN112687231A (en) | Brightness and chrominance data extraction method, equipment and computer readable storage medium | |
| CN118014832B (en) | Image stitching method and related device based on linear feature invariance | |
| US9554121B2 (en) | 3D scanning apparatus and method using lighting based on smart phone | |
| CN102496147A (en) | Image processing device, image processing method and image processing system | |
| CN114332183B (en) | Image registration method, device, computer equipment and storage medium | |
| CN113313707A (en) | Original image processing method, device, equipment and readable storage medium | |
| JP2012050013A (en) | Imaging apparatus, image processing device, image processing method, and image processing program | |
| US20040109599A1 (en) | Method for locating the center of a fiducial mark | |
| US20090021595A1 (en) | Low Memory Auto-Focus and Exposure System for Large Multi-Frame Image Acquisition | |
| JP2007074578A (en) | Image processing apparatus, photographing apparatus, and program | |
| CN105758337B (en) | A method of obtaining angle between lens plane and image sensor plane | |
| CN112637573A (en) | Multi-lens switching display method and system, intelligent terminal and storage medium | |
| US9489727B2 (en) | Method for generating a preferred image by replacing a region of a base image | |
| CN107644442B (en) | Spatial position calibration method of double-camera module | |
| CN116883240A (en) | Image stitching method based on gray level matching | |
| CN108133493A (en) | A kind of heterologous image registration optimization method mapped based on region division and gradual change | |
| US20170161927A1 (en) | Systems and Methods for Stitching Metallographic and Stereoscopic Images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190205 Termination date: 20210612 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |