CN101930606A - Field depth extending method for image edge detection - Google Patents
Field depth extending method for image edge detection Download PDFInfo
- Publication number
- CN101930606A CN101930606A CN 201010176120 CN201010176120A CN101930606A CN 101930606 A CN101930606 A CN 101930606A CN 201010176120 CN201010176120 CN 201010176120 CN 201010176120 A CN201010176120 A CN 201010176120A CN 101930606 A CN101930606 A CN 101930606A
- Authority
- CN
- China
- Prior art keywords
- image
- gray level
- cloth
- width
- edge detection
- 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
Images
Abstract
The invention discloses a field depth extending method for image edge detection, which is used for image processing. The method comprises the following steps of: scanning along the depth (height) direction of a photographed object with the lens of an auto-focus microscope and automatically sampling and photographing the object once every a certain distance; performing Sobel operator edge detection on one image in real time every time when the image is photographed so as to obtain a gray level image only containing the edge of a clear part; performing expansion transformation on each image which is subjected to edge detection so as to extend and connect the clear edge part of the gray level image into a series of connected regions; and comparing the gray level image obtained after the expansion transformation with the previous grey level image point by point, retaining the initial image value at the point of the image with the larger gray level value, combining the values to form a new image serving as an intermediate result, and continuously combining the new image with a newly photographed image until the entire process is finished. The method is simpler than a wavelet transform method; and both the edge detection step and the morphologic transform step of the method have small calculation amount, so that the speed advantage is showed.
Description
Technical field
The present invention relates to digital image processing techniques, relate in particular to a kind of field depth extending method of Image Edge-Detection.
Background technology
Depth of field expansion technique is a technology based on Digital Image Processing, the multiple image that it utilizes the associated picture information of gathering under the different scenes of same imageing sensor or different images sensor to take carries out image co-registration, with the multiple image information fusion in the middle of same width of cloth image.In the process of optical imagery, owing to be subjected to the restriction of the camera lens depth of field, piece image often can not well reflect the full detail of the subject with certain depth or height.And its depth of field of camera lens that enlargement factor is high is more little, that is to say that when we removed to take the object that certain depth or height arranged with this camera lens, we can only see its upper surface or lower surface clearly, and can not obtain width of cloth picture fully clearly.We just can take a series of pictures along the degree of depth (highly) direction of subject utilization depth of field expansion technique, use this method that it is synthesized then, thereby can obtain the synthetic deeply picture of a width of cloth panorama, reach the purpose of depth of field expansion.Depth of field expansion technique has become a kind of very important and useful graphical analysis and computer vision technique in recent years, field in automatic target identification, remote sensing, micrometering, Medical Image Processing and Military Application has a wide range of applications, existing as patent: ZL 03107895.8, and the biological microscope image technology reaches the method that increases the three-dimensional depth of field and resolution.
The simplest existing depth image synthetic method is that source images is weighted on average, and the advantage of this method is simple, and real-time is good, but the negative effect that meanwhile brings is the contrast that has reduced image, and depth of field synthetic effect is poor.The depth of field synthetic method of decomposing based on image pyramid also is many methods of research at present, and it has comprised that the wavelet transformation of image, how fast wave filter represent and pyramid transform.Small echo has represented a lot of superiority in graphical representation, its advantage is that the synthetic precision height of image is effective, and is slow but shortcoming also clearly is exactly a speed.Tens width of cloth image sequences synthesize piece image often in micrometering, if need spend a large amount of time with the wavelet transformation rule, this obviously is that the expansion of the restriction depth of field is tending towards the big bottleneck that product is used.Synthesize if merely adopt the method for pyramid transform to carry out the depth of field, but speed is to guarantee that synthetic effect is general that clear micro-image than higher is then not competent for requiring.
Therefore develop and a kind ofly can satisfy synthetic accuracy requirement, the Image Edge-Detection field depth extending method that can reach fast synthetic purpose again seems very necessary, is a technical matters that needs to be resolved hurrily in the industry.
Summary of the invention
The present invention is directed to the deficiency of existing depth of field synthetic method, a kind of both can satisfy synthetic precision of the depth of field and effect are provided, significantly improve the field depth extending method of the Image Edge-Detection of aggregate velocity again.
For solving the problems of the technologies described above, the field depth extending method of a kind of Image Edge-Detection that the present invention proposes, step is as follows:
Scan with the degree of depth (highly) direction of the microscopical camera lens of automatic focusing, and sample automatically at a certain distance and take once along subject;
Every shooting piece image carries out the Sobel operator edge detection to it in real time, obtains the gray level image that a width of cloth only contains the edge, clear area;
Image after each breadths edge detection is carried out dilation transformation, make the marginal portion expansion of the original clear area of gray level image be linked to be a series of connected regions;
The gray-scale map and the last width of cloth gray-scale map that obtain behind the dilation transformation are carried out pointwise relatively, and that width of cloth figure that gray-scale value is big remains in the image initial value of this point, and a synthetic width of cloth is newly schemed; This new figure is continued to synthesize with the new image of taking as intermediate result, finish up to whole process.
Wherein, described dilation transformation is the mathematical morphology dilation transformation.
Before the described automatic focusing microscope photographing, it is imported corresponding controlled variable in advance.
Described controlled variable can be spacing or time.
Compared with prior art, the present invention adopts that calculated amount is little, speed faster the Sobel edge detection operator sequence image that photographs is extracted marginal operation, utilize the result of rim detection to carry out mathematical image morphology dilation transformation again, thereby determine the clear area of each width of cloth image apace.Be different from Wavelet Transform, it at first needs image is carried out frequency domain transform, the similarity measure of computed image and conspicuousness are estimated again, and then estimate size according to conspicuousness and adopt suitable convergence strategy to synthesize, its wavelet transformation, similarity measure all are the bigger steps of calculated amount, and calculated amount is big along with the increase of image size is exponential change, in case image reaches the mega pixel level, its aggregate velocity will significantly descend.Therefore rim detection and the mathematical morphology dilation transformation among the present invention program all is the very little step of calculated amount, so its speed advantage has just displayed.Adopt the Sobel operator edge detection to have the fast advantage of aggregate velocity, therefore its synthesis step is embedded in the process of camera lens scanning shoot in real time, take the mode synthetic, embodied the fast advantage of this method aggregate velocity to greatest extent while clapping.The user in use can stop etc. scanning process along with the scanning discovery image of camera lens is more and more clear, then can obtain the image after a width of cloth panorama is expanded deeply at once.
Description of drawings
Below in conjunction with drawings and Examples the present invention is made detailed explanation, wherein:
Fig. 1 is the process flow diagram of preferred embodiments of the present invention;
Fig. 2 is the structural representation of image sequence figure of the present invention.
Embodiment
Preferred embodiments of the present invention as shown in Figure 1, the field depth extending method of described Image Edge-Detection has the following steps:
Step 1: scan with the degree of depth (highly) direction of microscopical camera lens of focusing automatically, and sample automatically at a certain distance and take once along subject.
Automatically focusing microscope is imported corresponding controlled variable in advance, as parameters such as the distance between the input control focusing microscope photographing image, times.The distance of sampling interval will determine the effect of depth of field expansion, also have influence on the precision that the degree of depth (highly) is measured simultaneously, and this sampled distance also is limited by the resolution of driving mechanism and the depth of field of camera lens simultaneously.In general, the resolution and the camera lens depth of field of driving mechanism are more little, and synthetic precision is high more, and the sequence of source images number is many more, and the scanning step of camera lens (distance) is more little when taking at every turn, and the degree of depth of measuring (highly) information is accurate more, and synthetic precision is high more.
Step 2: every shooting piece image, in real time it is carried out rim detection, obtain the gray level image that a width of cloth only contains the edge, clear area.
Wherein, rim detection adopts the Sobel operator edge detection.Suo Beier operator (Sobel operator) is one of operator relatively more commonly used in the Flame Image Process, mainly as rim detection.Technically, it is a discreteness difference operator, is used for the approximate value of gradient of arithmograph image brightness function.Its advantage is exactly that speed is fast, the precision height.Therefore, this method adopts the operator of Sobel operator as rim detection, purpose is to find clear part and fuzzy boundary partly in the source images, be beneficial to further the clear part and the fuzzy part of image border are made a distinction, obtain the gray level image that a width of cloth only contains clear part edge at last, the image that promptly obtains is that the marginal portion of former clear area has gray scale to show that the gray-scale value of other position (inside, clear area and fuzzy region) all is 0.When every shooting piece image, just in real time it is carried out the Sobel rim detection, obtain the gray level image that a width of cloth only contains clear part edge, wherein the size of edge gray-scale value has determined the readability that it is actual, be that the edge gray-scale value is high more, the sharpness of this position on former figure is high more.
Step 3: the image after each breadths edge detected carries out dilation transformation, the edge clear of the gray level image that above-mentioned steps 2 obtains is partly expanded be linked to be a series of connected regions.
Wherein, dilation transformation is the mathematical morphology dilation transformation.Result images after the rim detection is carried out the mathematical morphology dilation transformation, pass through dilation transformation, originally detected sharp edge can enlarge and connect into a slice integral body, that is to say by dilation transformation and just can be transformed into the zone of forming by face by the sharp edge that point or line are formed originally, be about to clearly or polar expansion is expanded, it is interconnected becomes a face district.So more help clear part among the former figure and fuzzy part separated, thereby successfully extract the zone of clear part in each width of cloth source images, can take clearly zone in the just current camera lens field depth.The image that promptly obtains is except the edge of clear area, and the inside of clear area also has gray scale to show, and the gray-scale value of other position (fuzzy region) still is 0.
Step 4: the back width of cloth gray-scale map and the last width of cloth that obtain behind the dilation transformation are carried out pointwise relatively through the gray-scale map behind the dilation transformation, that width of cloth figure that gray-scale value is big remains in the image initial value of this point, and synthesize a width of cloth and newly scheme, this new figure is exactly that a width of cloth has extracted front and back two width of cloth figure depth of field composite diagram of clear part separately.Image after synthetic is presented on the user interface in real time, and the new figure of this width of cloth continues to take the synthetic new figure of new image and intermediate result figure as intermediate result figure, finishes up to whole shooting process.
The image co-registration of this method (merging) process is: with the image unified Definition behind the mathematical morphology dilation transformation is I
n(x, y), wherein n represents the sequence number of the sequence image of all shootings, (x y) represents certain some position in image, and then (x y) is exactly the pixel value (gray-scale value) of this point to I.When merging, certain point of two width of cloth figure that front and back are adjacent compares, the sequence number n of that piece image that gray-scale value is bigger notes, and its sequence number is filled into the position of this point, and (x y), has so just formed the new image of a width of cloth, only this width of cloth image is not real image, because what write down in each pixel on this width of cloth image is not the gray-scale value of image, but the sequence number n of image, as shown in Figure 2.The purpose that produces this width of cloth sequence number image is exactly when final image merges, and system goes up the sequence number n of record automatically according to each point in this width of cloth image, decides the pixel value of this point should adopt the pixel value of the corresponding point of which width of cloth figure in the original image sequence.
Is that example describes in detail with Fig. 2, after the conversion through the front, the gray-scale value of each point of image sequence is compared draw the such sequence chart of Fig. 2.If the size of image is (indulging) 10 * (horizontal stroke) 15, have 10 width of cloth source images and participate in merging, sequence number n is labeled as 0,1,2 respectively ... 9.According to sequence chart, the data that (0,0) is located are 4, then can draw, and fused images (0,0) is located, the pixel value that should select the 4th width of cloth source images (0,0) to locate; In like manner the data of (9,11) position are 8 in the sequence chart, and fused images (9,11) is located the pixel value that just should select the 8th width of cloth source images (9,11) to locate so.Like this, the guide according to sequence of source images in the data template number just can obtain the pixel value that all pixels of entire image should be chosen in fusion process.
Can also measure the degree of depth (highly) information of shot object by this method, to focusing microscope, the user can be provided with the motor-driven camera lens and scan by fixed step size along the degree of depth (highly) direction of subject by automatically.Will take the source images of one group of 10 width of cloth sequence such as the user, every shooting piece image camera lens moves 1 μ m, and sequence number is that 3 location points and sequence number are that the degree of depth (highly) difference between 9 the location point is exactly (9-3) * 1=6 μ m in sequence chart so.The sequence of source images number is many more, and the scanning step of camera lens is more little when taking at every turn, and the measured degree of depth (highly) value of coming out is just accurate more.
The comparison on processing speed with this method and small wave converting method, as following table:
| The image size | 200×200 | 640×480 | 1344×1024 | |
| Ask during processing (unit: second) | This method | 0.132 | 0.414 | 0.832 |
| Wavelet transformation | 1.341 | 5.256 | 20.442 |
By means of motor-driven camera lens, camera can carry out interval shooting along the subject degree of depth (highly) conversion direction, synthesizes and real-time display result by the depth of field that this method computing machine can be finished in each interval of taking each time fully.When scanning process finished, the composograph after the dark expansion of final panorama also disposed fully, and simultaneously, according to the step-length of shooting interval each time, the user can calculate the degree of depth (highly) of captured object.This method is very obvious than the advantage of existing small wave converting method on processing speed, has actual application background, and its degree of depth (highly) measurement function has also further enriched the application prospect of this method simultaneously.
Adopt that calculated amount is little, speed faster the Sobel edge detection operator sequence image that photographs is extracted marginal operation, the result to rim detection carries out the mathematical morphology dilation transformation again, thereby determines the clear area of each width of cloth image apace.Synthetic different with general Wavelet Transform, wavelet transformation, similarity measure calculated amount are bigger, and when being used for the mega pixel level, its aggregate velocity will significantly descend.Rim detection among the present invention program and dilation transformation calculated amount are very little by contrast, and its speed advantage has just displayed.
Abovely the present invention is specifically described, but those skilled in the art can make numerous variations or variation to these embodiments these changes and change and to fall within the scope of protection of the invention in conjunction with better embodiment.
Claims (4)
1. the field depth extending method of an Image Edge-Detection is characterized in that, comprises the steps:
Scan with the depth direction of the microscopical camera lens of automatic focusing, and sample automatically at a certain distance and take once along subject;
Every shooting piece image carries out the Sobel operator edge detection to it in real time, obtains the gray level image that a width of cloth only contains the edge, clear area;
Gray level image after each breadths edge detection is carried out dilation transformation, make the marginal portion expansion of the original clear area of gray level image be linked to be a series of connected regions;
The gray level image and the last width of cloth gray level image that obtain behind the dilation transformation are carried out pointwise relatively, and that width of cloth figure that gray-scale value is big remains in the image initial value of this point, and a synthetic width of cloth is newly schemed; This new figure is continued to synthesize with the new image of taking as intermediate result, finish up to whole process.
2. field depth extending method according to claim 1 is characterized in that: described dilation transformation is the mathematical morphology dilation transformation.
3. field depth extending method according to claim 2 is characterized in that, before the described automatic focusing microscope photographing, it is imported the control corresponding parameter in advance.
4. field depth extending method according to claim 3 is characterized in that, described controlled variable is spacing, time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010176120 CN101930606A (en) | 2010-05-14 | 2010-05-14 | Field depth extending method for image edge detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010176120 CN101930606A (en) | 2010-05-14 | 2010-05-14 | Field depth extending method for image edge detection |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101930606A true CN101930606A (en) | 2010-12-29 |
Family
ID=43369765
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010176120 Pending CN101930606A (en) | 2010-05-14 | 2010-05-14 | Field depth extending method for image edge detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101930606A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102957871A (en) * | 2012-11-06 | 2013-03-06 | 广东欧珀移动通信有限公司 | Field depth processing method for shot image of mobile terminal |
| CN104007047A (en) * | 2014-06-17 | 2014-08-27 | 青岛理工大学 | Method for identifying granular system power transmission chain |
| CN104639831A (en) * | 2015-01-05 | 2015-05-20 | 信利光电股份有限公司 | Camera and depth of field expanding method |
| CN104660900A (en) * | 2013-10-30 | 2015-05-27 | 株式会社摩如富 | Image Processing Device, Image Processing Method And Recording Medium |
| CN110764244A (en) * | 2019-11-05 | 2020-02-07 | 安图实验仪器(郑州)有限公司 | Automatic focusing method for microscope tabletting microscopic examination |
| CN111242880A (en) * | 2019-12-30 | 2020-06-05 | 广州市明美光电技术有限公司 | Multi-depth-of-field image superposition method, equipment and medium for microscope |
| CN111856739A (en) * | 2019-04-30 | 2020-10-30 | 莱卡微系统Cms有限责任公司 | Image processing apparatus, system and method for improving signal-to-noise ratio of microscope images |
| CN112763466A (en) * | 2020-12-24 | 2021-05-07 | 山东省交通科学研究院 | Method for identifying phase state distribution characteristics of polymer modified asphalt |
| CN113420780A (en) * | 2021-05-24 | 2021-09-21 | 中国科学院地理科学与资源研究所 | Culture pond extraction method based on remote sensing spatial-temporal spectral feature fusion |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1534323A (en) * | 2003-04-01 | 2004-10-06 | 翰沃生电科技股份有限公司 | Method of making biological microscope imaging technique reach increasing 3D field depth and resolution degree |
| CN101021945A (en) * | 2007-03-23 | 2007-08-22 | 北京中星微电子有限公司 | Image composing method and device |
| CN101322418A (en) * | 2005-12-02 | 2008-12-10 | 皇家飞利浦电子股份有限公司 | Depth dependent filtering of image signal |
| CN101640809A (en) * | 2009-08-17 | 2010-02-03 | 浙江大学 | Depth extraction method of merging motion information and geometric information |
-
2010
- 2010-05-14 CN CN 201010176120 patent/CN101930606A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1534323A (en) * | 2003-04-01 | 2004-10-06 | 翰沃生电科技股份有限公司 | Method of making biological microscope imaging technique reach increasing 3D field depth and resolution degree |
| CN101322418A (en) * | 2005-12-02 | 2008-12-10 | 皇家飞利浦电子股份有限公司 | Depth dependent filtering of image signal |
| CN101021945A (en) * | 2007-03-23 | 2007-08-22 | 北京中星微电子有限公司 | Image composing method and device |
| CN101640809A (en) * | 2009-08-17 | 2010-02-03 | 浙江大学 | Depth extraction method of merging motion information and geometric information |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102957871A (en) * | 2012-11-06 | 2013-03-06 | 广东欧珀移动通信有限公司 | Field depth processing method for shot image of mobile terminal |
| US10127639B2 (en) | 2013-10-30 | 2018-11-13 | Morpho, Inc. | Image processing device having depth map generating unit, image processing method and non-transitory computer readable recording medium |
| CN104660900A (en) * | 2013-10-30 | 2015-05-27 | 株式会社摩如富 | Image Processing Device, Image Processing Method And Recording Medium |
| CN104660900B (en) * | 2013-10-30 | 2018-03-02 | 株式会社摩如富 | Image processing apparatus and image processing method |
| CN104007047B (en) * | 2014-06-17 | 2016-01-27 | 青岛理工大学 | The dynamofluidal recognition methods of a kind of granular system |
| CN104007047A (en) * | 2014-06-17 | 2014-08-27 | 青岛理工大学 | Method for identifying granular system power transmission chain |
| CN104639831A (en) * | 2015-01-05 | 2015-05-20 | 信利光电股份有限公司 | Camera and depth of field expanding method |
| CN104639831B (en) * | 2015-01-05 | 2018-12-11 | 信利光电股份有限公司 | A kind of camera and the method for expanding the depth of field |
| CN111856739A (en) * | 2019-04-30 | 2020-10-30 | 莱卡微系统Cms有限责任公司 | Image processing apparatus, system and method for improving signal-to-noise ratio of microscope images |
| CN110764244A (en) * | 2019-11-05 | 2020-02-07 | 安图实验仪器(郑州)有限公司 | Automatic focusing method for microscope tabletting microscopic examination |
| CN111242880A (en) * | 2019-12-30 | 2020-06-05 | 广州市明美光电技术有限公司 | Multi-depth-of-field image superposition method, equipment and medium for microscope |
| CN112763466A (en) * | 2020-12-24 | 2021-05-07 | 山东省交通科学研究院 | Method for identifying phase state distribution characteristics of polymer modified asphalt |
| CN113420780A (en) * | 2021-05-24 | 2021-09-21 | 中国科学院地理科学与资源研究所 | Culture pond extraction method based on remote sensing spatial-temporal spectral feature fusion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101930606A (en) | Field depth extending method for image edge detection | |
| US8787656B2 (en) | Method and apparatus for feature-based stereo matching | |
| CN104574393B (en) | A kind of three-dimensional pavement crack pattern picture generates system and method | |
| CN105279372B (en) | A kind of method and apparatus of determining depth of building | |
| CN107993258B (en) | Image registration method and device | |
| CN107687819B (en) | A kind of optical losses sub-pixel extraction of quick high accuracy | |
| CN103793918A (en) | Image definition detecting method and device | |
| CN105930822A (en) | Human face snapshot method and system | |
| CN103714321B (en) | Driver's Face detection system based on range image and intensity image | |
| CN109360235A (en) | A kind of interacting depth estimation method based on light field data | |
| CN104101294B (en) | Analytical equipment, analysis program and analysis system | |
| CN111160291B (en) | Human eye detection method based on depth information and CNN | |
| CN110736747B (en) | Method and system for positioning under cell liquid-based smear mirror | |
| KR20090037275A (en) | Detecting apparatus of human component and method of the same | |
| CN101957915A (en) | Calligraphy and painting scanning equipment and identification method | |
| Song et al. | Combining stereo and time-of-flight images with application to automatic plant phenotyping | |
| CN106504274A (en) | A kind of visual tracking method and system based under infrared camera | |
| CN115115672B (en) | Dynamic vision SLAM method based on target detection and feature point speed constraint | |
| CN103700082B (en) | Image split-joint method based on dual quaterion relative orientation | |
| CN112396073A (en) | Model training method and device based on binocular images and data processing equipment | |
| CN111292228A (en) | Lens defect detection method | |
| CN102682435B (en) | Multi-focus image edge detection method based on space relative altitude information | |
| US9995905B2 (en) | Method for creating a camera capture effect from user space in a camera capture system | |
| CN105809664A (en) | Method and device for generating three-dimensional image | |
| CN106558069A (en) | A kind of method for tracking target and system based under video monitoring |
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: 20101229 |