CN109145910A - The image line feature extracting method of the reconstruction of hologram - Google Patents
The image line feature extracting method of the reconstruction of hologram Download PDFInfo
- Publication number
- CN109145910A CN109145910A CN201710504185.3A CN201710504185A CN109145910A CN 109145910 A CN109145910 A CN 109145910A CN 201710504185 A CN201710504185 A CN 201710504185A CN 109145910 A CN109145910 A CN 109145910A
- Authority
- CN
- China
- Prior art keywords
- beamlet
- image
- square
- dictionary
- hologram
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/28—Character recognition specially adapted to the type of the alphabet, e.g. Latin alphabet
- G06V30/287—Character recognition specially adapted to the type of the alphabet, e.g. Latin alphabet of Kanji, Hiragana or Katakana characters
Abstract
The image line feature extracting method of the reconstruction of hologram proposed by the present invention, is related to hologram image processing technology field, and 1, for bianry image after pretreatment, edge detection is done with Canny operator;2, the Beamlet dictionary that scale size is 16 × 16,8 × 8 and 4 × 4 is established respectively;3, be 256 × 256 by 1 result images size adjusting, by image by two into model split at the son square of 16 × 16 sizes, and 16 × 16 dictionary established using step 2 calculates the Beamlet coefficient of each height square on the scale;4, the maximum value on each height square is found, a threshold value T is then set, when maximum value is greater than T, then crosses and shows this Beamlet;5,3 and 4 are repeated in using 8 × 8 and 4 × 4 dictionary respectively.The present invention can both detect the contour edge of Chinese character in image by Canny operator, it can use Beamlet again and detect the fine edge with good directivity, the pseudo-edge that may detect that when removing Canny operator edge detection simultaneously obtains Chinese character contour as accurate as possible.
Description
Technical field
The present invention relates to the image line feature extracting methods of hologram image processing technology field, the especially reconstruction of hologram.
Background technique
Image line is characterized in the important feature in the important clue and pattern-recognition of visual perception, corresponds to image spy
The place that property changes.Therefore, the important step that the line in image is characterized in image procossing and computer vision is extracted, it
Direct basis is provided for image recognition.In general, in image there are three types of the modes of line feature extraction: the first is to pass through line template
Matching is to detect the straight line on specific direction;Second is directly to carry out edge detection to image using edge detection operator to obtain
Line feature;The third is using the method for transformation, as Hough transform extracts the line feature in image.
Summary of the invention
The image line feature extracting method of the reconstruction of hologram provided by the invention, convenient for calculating suitable Hanzi features for Chinese character
Identification.
The present invention specifically adopts the following technical scheme that realization:
A kind of image line feature extracting method of the reconstruction of hologram, specifically includes the following steps:
Step 1 is directed to bianry image after pretreatment, as shown in Fig. 2, doing edge detection with Canny operator, wherein value is
1.5;
Step 2 establishes the Beamlet dictionary that scale size is 16 × 16,8 × 8 and 4 × 4 respectively, and δ is that a pixel is big
It is small, a point is respectively chosen on any both sides of square, and connect to get to a Beamlet, record on each Beam
Location information, lnAnd L;
Step 3, by the result images size adjusting of step 1 be 256 × 256, by image by two into model split at
The son square of 16 × 16 sizes, and 16 × 16 dictionary established using step 2, it is square to calculate each height on the scale
The Beamlet coefficient of shape;
Step 4 calculates accordinglyAnd the maximum value on each height square is found, then set
One threshold value T then crosses when maximum value is greater than T and shows this Beamlet;
Step 5 uses 8 × 8 and 4 × 4 dictionary to be repeated in step 3 and step 4 respectively.
The image line feature extracting method of the reconstruction of hologram provided by the invention, the beneficial effect is that: it is first to be calculated with Canny
Son carries out edge detection to bianry image, then with extracting local line feature based on the Beamlet line detection algorithms of single scale.
The advantage of doing so is that Canny operator is combined with Beamlet transformation, can both be detected in image by Canny operator
The contour edge of Chinese character, and can use Beamlet and detect the fine edge with good directivity, while removing Canny
The pseudo-edge that may detect that when operator edge detection, it is final to obtain Chinese character contour as accurate as possible, it is suitable convenient for calculating
Hanzi features are for Chinese Character Recognition.
Detailed description of the invention
Fig. 1 is the flow chart of image line feature extracting method of the present invention;
Fig. 2 is bianry image schematic diagrames after pretreatment;
Fig. 3 is the Beamlet schematic diagram of different scale;
Fig. 4, which is multiple dimensioned Beamlet, approaches schematic diagram to any one line segment;
Fig. 5 is Canny operator edge detection result schematic diagram;
Fig. 6 is the Beamlet testing result schematic diagram having a size of 16 × 16;
Fig. 7 is having a size of 8 × 8Beamlet testing result schematic diagram;
Fig. 8 is the Beamlet testing result schematic diagram having a size of 4 × 4.
Specific embodiment
To further illustrate that each embodiment, the present invention are provided with attached drawing.These attached drawings are that the invention discloses one of content
Point, mainly to illustrate embodiment, and the associated description of specification can be cooperated to explain the operation principles of embodiment.Cooperation ginseng
These contents are examined, those of ordinary skill in the art will be understood that other possible embodiments and advantages of the present invention.In figure
Component be not necessarily to scale, and similar component symbol is conventionally used to indicate similar component.
Now in conjunction with the drawings and specific embodiments, the present invention is further described.
A kind of image line feature extracting method of reconstruction of hologram provided in this embodiment is first with Canny operator to two-value
Image carries out edge detection, then with extracting local line feature based on the Beamlet line detection algorithms of single scale.It does so
Benefit is to combine Canny operator with Beamlet transformation, the wheel of Chinese character in image can be both detected by Canny operator
Wide edge, and can use Beamlet and detect the fine edge with good directivity, while removing Canny operator edge
The pseudo-edge that may detect that when detection, it is final to obtain Chinese character contour as accurate as possible, convenient for calculating suitable Hanzi features
For Chinese Character Recognition.As shown in Figure 1, specific method and step is as follows:
Step 1 is directed to bianry image after pretreatment, as shown in Fig. 2, doing edge detection with Canny operator, wherein value is
1.5;
Step 2 establishes the Beamlet dictionary that scale size is 16 × 16,8 × 8 and 4 × 4 respectively, and δ is that a pixel is big
It is small, a point is respectively chosen on any both sides of square, and connect to get to a Beamlet, record on each Beam
Location information, lnAnd L;
Step 3, by the result images size adjusting of step 1 be 256 × 256, by image by two into model split at
The son square of 16 × 16 sizes, and 16 × 16 dictionary established using step 2, it is square to calculate each height on the scale
The Beamlet coefficient of shape;
Step 4 calculates accordinglyAnd the maximum value on each height square is found, then set
One threshold value T then crosses when maximum value is greater than T and shows this Beamlet;
Step 5 uses 8 × 8 and 4 × 4 dictionary to be repeated in step 3 and step 4 respectively.
In step 1, the basic principle of Canny operator edge detection is: upward using either two-dimensional Gaussian function
First directional derivative is filtered image as filter, the maximum of gradient in calculated result image, so that it is determined that image
Edge.Canny operator is by calculating gradient come when determining whether pixel is marginal point, it is also contemplated that the shadows of other pixels
It rings, it is judged jointly according to current pixel and processed pixel.Canny operator has unique performance,
The effect of edge detection is obvious.
In step 2, Beamlet is counted as the extension of small echo (Wavelet), because it has possessed by small echo more points
Resolution frame is similarly a kind of tool of multi-scale geometric analysis.The line segment that any two point is connected in image is known as Beam,
About O (n4) item.In order to reduce radix, thus the concept of Beamlet is introduced, Beamlet is defined by following three step:
(1) recurrence two is into piecemeal
Assuming that image is unit square, 2 × 2 equal-sized son squares are divided into, then just every height
It is rectangular to divide 2 × 2 smaller equal-sized son squares.This operation is repeated, until being divided into optimal scale.
(2) apex marker
Since the vertex in the upper left corner of every height square, with an equal distance δ according to (or inverse time clockwise
Needle) bearing mark square side.Wherein, δ is constant selected in advance.
(3) vertex is connected
In each height square, the mark determined in any pair of step (2) on the square side is connected with straight line
Remember that point, line segment as each are known as Beamlet.For the digital picture of n × n, there are about O (n2log2N) Beamlet.
Shown respectively in Fig. 3 it is in unit square and by unit square be divided into it is 2 × 2,4 × 4,8 × 8 small square when
An arbitrary Beamlet.
Beamlet dictionary be one with different location, direction and size two into the line segment library of composition, it can be used to
Any one line segment of multi-scale Retinex.Multiple dimensioned Beamlet chain approaches as shown in Figure 4 line segment.
Beamlet transformation for single scale, all windows size having the same, the structure of Beamlet also phase
Together.Therefore, Beamlet dictionary only needs to calculate once, so that it may suitable for all windows.For each Beamlet, need
Record following information:
(1) location information of the pixel on Beam;
(2) the line segment length l of the divided Beam of each windown;
(3) the length L of Beamlet;
(4) by (2) and (3), the weight of corresponding pixel can be calculated.
After information above is saved, Beamlet dictionary is just established.After image is divided into square wicket,
Each wicket can reuse the dictionary built up and convert to carry out Beamlet.
It is as shown in Figure 5 to obtain result by Canny operator edge detection according to the algorithm of the application.Then it uses
Beamlet extracts line feature and it is as shown in Figure 6 to 8 to respectively obtain result using 16 × 16,8 × 8 and 4 × 4 dictionaries.
Canny operator can preferably extract Chinese glyph, but some tiny noises are still had in result.Point
Not Cai Yong the Beamlet of 16 × 16,8 × 8 and 4 × 4 scales the result of Canny edge detection is continued to do line detection.From result
As can be seen that being limited by the design feature of Chinese character itself, when Beamlet scale is bigger, the line segment number of extraction is fewer, noise
Remove cleaner, but it is poorer to the fitting effect of Chinese glyph, there is the phenomenon that stroke outlines are lost or misplace;When
Beamlet scale is got over hour, and the line segment number of extraction is more, and the noise of removal is few, but the fitting to Chinese glyph
Effect is better.Therefore, by comprehensively considering, the result images that scale is 44 are chosen herein and do further Chinese character segmentation, are obtained
Chinese character pixel image to be identified.
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 (1)
1. a kind of image line feature extracting method of reconstruction of hologram, which is characterized in that specifically includes the following steps:
Step 1 is directed to bianry image after pretreatment, as shown in Fig. 2, doing edge detection with Canny operator, wherein value is 1.5;
Step 2 establishes the Beamlet dictionary that scale size is 16 × 16,8 × 8 and 4 × 4 respectively, and δ is a pixel size,
A point is respectively chosen on any both sides of square, and is connected to get to a Beamlet, and the position on each Beam is recorded
Confidence breath, lnAnd L;
The result images size adjusting of step 1 is 256 × 256 by step 3, by image by two into model split at 16 ×
The son square of 16 sizes, and 16 × 16 dictionary established using step 2 calculate each height square on the scale
Beamlet coefficient;
Step 4 calculates accordinglyAnd the maximum value on each height square is found, then set one
Threshold value T then crosses when maximum value is greater than T and shows this Beamlet;
Step 5 uses 8 × 8 and 4 × 4 dictionary to be repeated in step 3 and step 4 respectively.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710504185.3A CN109145910A (en) | 2017-06-28 | 2017-06-28 | The image line feature extracting method of the reconstruction of hologram |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710504185.3A CN109145910A (en) | 2017-06-28 | 2017-06-28 | The image line feature extracting method of the reconstruction of hologram |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109145910A true CN109145910A (en) | 2019-01-04 |
Family
ID=64805449
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710504185.3A Pending CN109145910A (en) | 2017-06-28 | 2017-06-28 | The image line feature extracting method of the reconstruction of hologram |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109145910A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112307803A (en) * | 2019-07-25 | 2021-02-02 | 中国石油天然气股份有限公司 | Digital geological outcrop crack extraction method and device |
-
2017
- 2017-06-28 CN CN201710504185.3A patent/CN109145910A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112307803A (en) * | 2019-07-25 | 2021-02-02 | 中国石油天然气股份有限公司 | Digital geological outcrop crack extraction method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2017219391A1 (en) | Face recognition system based on three-dimensional data | |
CN108694393A (en) | A kind of certificate image text area extraction method based on depth convolution | |
WO2006087580A1 (en) | Fast method of object detection by statistical template matching | |
CN105447512A (en) | Coarse-fine optical surface defect detection method and coarse-fine optical surface defect detection device | |
CN110222661B (en) | Feature extraction method for moving target identification and tracking | |
WO2015066984A1 (en) | Complex background-oriented optical character recognition method and device | |
CN112734729B (en) | Water gauge water level line image detection method and device suitable for night light supplement condition and storage medium | |
CN108090492B (en) | Contour detection method based on scale clue suppression | |
CN112991374A (en) | Canny algorithm-based edge enhancement method, device, equipment and storage medium | |
CN108932492A (en) | A kind of image fingerprint extracting method based on non-sampled shearing wave conversion | |
CN109767442B (en) | Remote sensing image airplane target detection method based on rotation invariant features | |
CN101430789B (en) | Image edge detection method based on Fast Slant Stack transformation | |
CN108805139A (en) | A kind of image similarity computational methods based on frequency-domain visual significance analysis | |
CN109145910A (en) | The image line feature extracting method of the reconstruction of hologram | |
CN112001448A (en) | Method for detecting small objects with regular shapes | |
Cheng et al. | Power pole detection based on graph cut | |
CN106355576A (en) | SAR image registration method based on MRF image segmentation algorithm | |
CN113420633B (en) | Traffic sign identification method based on UM enhancement and SIFT feature extraction | |
CN111274858A (en) | Business license identification method in network transaction supervision | |
Rajan et al. | Text detection and character extraction in natural scene images using fractional Poisson model | |
OUYANG et al. | Edge detection in pavement crack image with beamlet transform | |
CN109086738A (en) | A kind of character identifying method and device based on template matching | |
Meng et al. | Comparative studies of multiscale edge detection using different edge detectors for MRI thigh | |
Ye et al. | Reading labels of cylinder objects for blind persons | |
CN114529715A (en) | Image identification method and system based on edge extraction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190104 |