CN100543576C - Based on the digital image-forming of the Contourlet conversion method of focusing automatically - Google Patents
Based on the digital image-forming of the Contourlet conversion method of focusing automatically Download PDFInfo
- Publication number
- CN100543576C CN100543576C CNB2008100922973A CN200810092297A CN100543576C CN 100543576 C CN100543576 C CN 100543576C CN B2008100922973 A CNB2008100922973 A CN B2008100922973A CN 200810092297 A CN200810092297 A CN 200810092297A CN 100543576 C CN100543576 C CN 100543576C
- Authority
- CN
- China
- Prior art keywords
- focusing
- image
- subband
- significant
- evaluation function
- 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
Images
Abstract
The invention discloses a kind of digital image-forming method of focusing automatically based on the Contourlet conversion, the multi-direction characteristic of multiresolution of having utilized the Contourlet conversion to be had, choose can the token image contour acuity and the significant coefficient in the Contourlet territory of details richness calculate focusing evaluation function, thereby determine the accurate focusing position of digital imaging system, for the image that comprises the texture that moves towards more consistent in a large number, utilize the dominant direction at texture edge to determine that the method for Contourlet territory dominant direction subband then can utilize the directional information of image better, thereby improve focusing sensitivity, and reduce computation complexity.
Description
The application is that application number is dividing an application of No. 200610053118.6 applications for a patent for invention.
Technical field
The present invention relates to the digital image-forming method of focusing automatically, especially relate to a kind of digital image-forming method of focusing automatically based on the Contourlet conversion.
Background technology
In imaging system, camera lens has a best image planes position to object image-forming, departs from this position and will cause image blurringly, and image quality descends; Can therefore, accurately focus to an imaging system be crucial.Imaging system based on digital picture adopts automatic focusing method, and its key is focusing evaluation function.Desirable focusing evaluation function curve shows as the class parabolic shape, and its peak value is corresponding to the optimal imaging position, and functional value reduces when leaving optimum.Therefore automatic focusing process comes down to ask for the peaked process of focusing evaluation function.
Usually, the energy major part of image concentrates on the low frequency and the Mid Frequency of frequency domain, but the richness of the acutance of image outline and details then depends on the high frequency composition of image.When clear picture, details is abundant, and eigenwert (as gray scale, the color etc.) variation that shows as neighbor on the spatial domain is bigger, and the high fdrequency component that then shows as frequency spectrum at frequency domain is many.Focusing evaluation function commonly used is divided into two kinds: the spatial domain with frequency domain.Several spatial domains focusing evaluation function commonly used comprises laplacian operator, sobel operator, prewitt operator and energy variance operator etc.Less relatively based on the operand that the focusing evaluation method in spatial domain is required, but its shortcoming is to be subjected to The noise bigger, and promptly noise immunity is relatively poor.Frequency domain focusing evaluation method has then been utilized the overall permanence of image, and noise immunity is better relatively, but this method needs elder generation that image is carried out Fourier transform or other conversion usually, comes the sharpness of evaluation map picture again according to conversion coefficient.
The multi-direction multiple dimensioned characteristic of Contourlet conversion makes it can catch texture, details and marginal information in the natural image well, and the automatic focusing process of digital picture estimates and asks the process of extreme point just to the sharpness of information such as grain details, therefore, the Contourlet conversion is a kind of instrument that is fit to focus automatically and estimates.Numeral Contourlet conversion is divided into two steps, at first use the tower bank of filters of Laplce (Laplacian Pyramid, LP) image being carried out multiresolution decomposes and catches singular point, (Directional FilterBank, DFB) that the position is close singular point accumulates contour segment according to its different directivity characteristics to use the two-dimensional directional bank of filters then.The Contourlet coefficient of each grade has a plurality of direction subbands, and these subbands are the edge image that contains high fdrequency component.After image being carried out one-level Contourlet decomposition, can obtain the respective edges characteristic image.Be Fig. 1 (a) as Fig. 1 (b) and carry out 2 grades of Contourlet conversion coefficient figure after the Contourlet conversion.But for imaging system, this has just obtained a discrete edge feature image sequence, realize accurate focusing, also must set up an objective function, and discrete picture is provided statistical property, is used for judging best focusing position.The process of focusing is the process that an image energy changes, and the image of general accurate focusing has maximum energy.
Summary of the invention
Technical matters to be solved by this invention provides a kind of accurately and effectively based on the digital image-forming of the Contourlet conversion method of focusing automatically.
The present invention solves the problems of the technologies described above the technical scheme that is adopted: a kind of digital image-forming based on Contourlet conversion method of focusing automatically, it may further comprise the steps: 1. for the digital micro imaging system, continuously focusing is with the input signal of the image that obtains sharpness and have nothing in common with each other; 2. the two-dimentional input signal f to every width of cloth image carries out multistage Contourlet conversion, obtains Contourlet set of transform coefficients { C
J, k, C here
J, kExpression Contourelet territory j layer (j=1 ..., J, J are the highest layer of resolution) matrix of coefficients of a k subband; 3. analyze definite important directions that needs the image of focusing according to the energy size of all directions subband, make the subband that comprises important directions information in focusing evaluation function, account for leading composition, comprise the less subband of directional information and account for less share, and determine the importance of different directions subband by weighting scheme, promptly define focusing evaluation function and be:
K wherein
jBe the sub band number of j layer, w
J, kBe weights,
The energy summation of significant coefficient in k subband of expression Contourelet territory j layer, promptly
Or
Be the significant coefficient in k subband of Contourelet territory j layer, promptly
N is the number of the significant coefficient chosen in this subband, the input signal of every width of cloth image is calculated the value of its focusing evaluation function F; 4. when the variation tendency of focusing back focusing evaluation function F value changed continuously, the readjustment focal length occurred until the image input signal corresponding to focusing evaluation function F maximum value, finishes the focusing process then; 5. get the image that promptly obtains sharpness the best of correct focusing corresponding to the image of the focusing evaluation function F maximum value of input signal.
Described important directions comprises dominant direction, can only adopt the significant coefficient of the determined subband of dominant direction to calculate focusing evaluation function.
The Zone Full that two-dimentional input signal f in the said method can be the digital micro imaging image or the combination or the down-sampled signal of regional area or regional area.
Compare with existing classical focusing evaluation method, digital image-forming based on the Contourlet conversion method of focusing automatically of the present invention has been utilized Contourlet conversion image orientation information is caught and possible direction number purpose dirigibility well, can extract the high-frequency information of image effectively, judge the image that has ceiling capacity in the image sequence, thereby determine the accurate focusing position of digital imaging system.
Description of drawings
Fig. 1 (a) is the capsicum original image;
Fig. 1 (b) is 2 grades of Contourlet coefficient in transform domain figure of capsicum;
Fig. 2 (a) is the microscope picture of the fuzzy ciliated epithelial cell section of focusing;
Fig. 2 (b) is the microscope picture of the ciliated epithelial cell section clearly of focusing;
Fig. 3 is the numbering figure of the top all directions subband of the resolution of 2 grades of Contourlet conversion;
Fig. 4 is the contrast based on the automatic focusing method of Contourlet conversion of selecting the different sub-band coefficient for use;
Fig. 5 is based on the automatic focusing method of Contourlet conversion and the contrast of classical focusing method.
Embodiment
Embodiment describes in further detail the present invention below in conjunction with accompanying drawing.
A kind of digital image-forming method of focusing automatically based on the Contourlet conversion, it may further comprise the steps: 1. for the digital micro imaging system, focusing is with the input signal of the image that obtains sharpness and have nothing in common with each other continuously; 2. the two-dimentional input signal f to every width of cloth image carries out multistage Contourlet conversion, obtains Contourlet set of transform coefficients { C
J, k, C here
J, kExpression Contourelet territory j layer (j=1 ..., J, J are the highest layer of resolution) matrix of coefficients of a k subband; 3. the Contourlet conversion has sparse property, promptly only carries out inverse transformation with a spot of coefficient in transform domain and just can approach original image effectively.With regard to Contourlet conversion high frequency coefficient, its amplitude overwhelming majority accumulates near zero, only have the amplitude of minute quantity coefficient bigger, the coefficient that this part amplitude is bigger is referred to as " significant coefficient " (significant), has described the detail textures information of image.Therefore, the evaluation of focusing of the significant coefficient of choosing some in can the different directions subband under different resolution promptly defines focusing evaluation function:
Wherein,
The energy summation of significant coefficient in k subband of expression Contourelet territory j layer, promptly
Be the significant coefficient in k subband of Contourelet territory j layer, promptly
N is the number of the significant coefficient chosen in this subband, the input signal of every width of cloth image is calculated the value of its focusing evaluation function F; 4. when the continuous variation tendency of focusing back focusing evaluation function F value changed (becoming successively decreases or become by successively decreasing increases progressively by increasing progressively), the readjustment focal length finished the focusing process then until the image input signal appearance corresponding to focusing evaluation function F maximum value; 5. get the image that promptly obtains sharpness the best of correct focusing corresponding to the image of the focusing evaluation function F maximum value of input signal.
In order to reduce computation complexity, reach the purpose of rapid focus, among the above-mentioned focusing evaluation function F
Also can be defined as
The Zone Full that two-dimentional input signal f in the said method can be the digital micro imaging image or the combination or the down-sampled signal of regional area or regional area.
Compare with frequency domain evaluation methods such as wavelet transformation, Fourier transforms, the great advantage of Contourlet conversion is the seizure of the directional information of image and possible direction number purpose dirigibility.The direction number of sub-bands of wavelet transformation is constant, has only level, vertical and to three directions in angular direction.In the Contourlet transform domain, the number of direction subband is more, and the seizure of directional information is also more flexible.In order to utilize this multi-direction characteristic of Contourlet conversion better, can analyze definite several important directions that need the image of focusing according to the energy size of all directions subband, make the subband that comprises important directions information in focusing evaluation function, account for leading composition, comprise the less direction subband of directional information and then give less share, the different importance of this different directions subband can realize by weighting scheme.In addition, because the sparse property of Contourlet conversion, the measurement of above-mentioned direction sub belt energy can realize by the coefficient of selected part " important " only.Be that focusing evaluation function also can be defined as:
K wherein
jBe the sub band number of j layer, w
J, kBe weights.
In addition, choosing of the direction subband that participates in the calculating focusing evaluation function, also can analyze definite dominant direction that needs the image of focusing, only adopt the significant coefficient of the determined some subbands of dominant direction to carry out the calculating of focusing evaluation function according to the energy size of all directions subband.
Advance by amasthenic lens is equidistant, the step pitch of whenever advancing, the mode of taking piece image, present embodiment has been gathered the micro-image of 18 width of cloth " ciliated epithelial cell " section sheets, and size of images is 512 * 512.Picture quality experienced from fuzzy to clear again to fuzzy process, then should present the Changing Pattern that reduces again from small to large for the focusing evaluation function that adopts.Fig. 2 (a) has provided two different width of cloth images of focusing order of accuarcy with Fig. 2 (b), promptly than the microscopic result who blurs and " ciliated epithelial cell " cuts sheet more clearly.The texture marginal information that this image series comprises mainly is divided into two big classes: the one, and a large amount of either large or small border circular areas and round dot.Because the directional information of circle is uniformly distributed on each direction, therefore there is not certain specific dominant direction; The 2nd, the train of thought of striated, the trend of these train of thoughts is roughly consistent, near 45.0 ° of directions, show the Contourlet transform domain, then be that energy concentrates on the 5th the direction subband corresponding to the highest resolution layer of as shown in Figure 32 grades of Contourlet conversion relatively.Fig. 4 has provided the five kinds of focusing evaluation function curves based on the Contourlet conversion for image sequence shown in Figure 2, for relatively convenient, has carried out normalized processing, and these five kinds of methods are:
M1: promptly the top energy of going up all direction subbands of Contourlet territory resolution and;
M2: promptly the energy of the top significant coefficient of going up all direction subbands of Contourlet territory resolution and, got 1% coefficient of amplitude maximum on each direction subband here;
M3: promptly adopt the resolution top different directions subband of going up in Contourlet territory is asked for the mode of focusing evaluation function according to its importance different weights, for the present embodiment, adopt following focusing evaluation function:
Wherein k is the sequence number of direction subband, w
kBe weights, the pairing 5 work song bands of dominant direction account for the bigger weight of tool;
M4: only adopt the energy size of the pairing 5 work song bands of dominant direction to estimate the focusing situation, promptly focusing evaluation function is: F=‖ E
J, 5‖
2
M5: only choose in the pairing 5 work song bands of dominant direction significant coefficient ask energy and, promptly focusing evaluation function is:
Here chosen 1% coefficient of amplitude maximum in the 5 work song bands.
Fig. 4 shows that the focusing evaluation function curve of these five kinds of evaluation methods all is parabolic unimodal curve, and peak of function point all appears at the 9th width of cloth image place, accurate focusing that promptly should the series micro-image.The difference of function curve is the sharp degree difference of curve, and is in five focusing appraisal curves, the most precipitous with the curve of M5, peak of curve is more sharp-pointed, and the curve ratio of M1 and M2 is more approaching, and the peak value place is all more smooth, and the sharpness of M3 and M4 is then between the centre.Article five, the sharp-pointed degree of curve is followed successively by: M1 ≈ M2<M3<M4<M5.The normalized curve figure of five kinds of methods shows, automatic focusing method based on the Contourlet conversion is effective and feasible, especially for the image of the texture marginal information that comprises more specific direction, the coefficient of employing dominant direction subband not only can improve the sensitivity of automatic focusing, and, reduced computation complexity owing to only adopt the part coefficient.
Fig. 5 provides the comparison based on the automatic focusing method of Contourlet conversion and some classical focusing evaluation methods, and the focusing evaluation algorithms of employing comprises: Laplacian operator, Sobel operator, Prewitt operator, energy variance operator (standard) and wavelet transformation (wavelet) focusing evaluation algorithms.As seen from Figure 5, normalized curve all is rendered as unimodal parabolic shape, but the peak value of Sobel operator and Prewitt operator appears at the 10th width of cloth image, is not inconsistent with all the other methods, promptly fails accurately to focus.Remaining classical focusing method can both reach the purpose of accurate focusing, and the most precipitous based on the curve of the method M1 of Contourlet conversion and M2.
In sum, because the Contourlet conversion has the multi-direction characteristic of multiresolution, can extract the high-frequency information of image effectively based on the focusing evaluation function of Contourlet conversion, judge the image that has ceiling capacity in the image sequence, thereby determine the accurate focusing position of digital imaging system.For the image that comprises the texture that moves towards more consistent in a large number, utilize the dominant direction at texture edge to determine that the method for Contourlet territory dominant direction subband then can utilize the directional information of image better, thereby improve focusing sensitivity, reduce computation complexity.
Under the situation of the spirit and scope of the universal that does not deviate from claim and equal scope and limited, the example that the present invention is not limited to specific details and illustrates here and describe.
Claims (3)
1, a kind of digital image-forming method of focusing automatically based on the Contourlet conversion, it may further comprise the steps: 1. for the digital micro imaging system, focusing is with the input signal of the image that obtains sharpness and have nothing in common with each other continuously; 2. the two-dimentional input signal f to every width of cloth image carries out multistage Contourlet conversion, obtains Contourlet set of transform coefficients { C
J, k, C here
J, kThe matrix of coefficients of expression Contourelet territory j layer k subband, j=1 wherein ..., J, J are the highest layer of resolution; 3. analyze definite important directions that needs the image of focusing according to the energy size of all directions subband, make the subband that comprises important directions information in focusing evaluation function, account for leading composition, comprise the less subband of directional information and account for less share, and determine the importance of different directions subband by weighting scheme, promptly define focusing evaluation function and be:
K wherein
jBe the sub band number of j layer, w
J, kBe weights,
The energy summation of significant coefficient in k subband of expression Contourelet territory j layer, promptly
Or
Be the significant coefficient in k subband of Contourelet territory j layer, promptly
N is the number of the significant coefficient chosen in this subband, the input signal of every width of cloth image is calculated the value of its focusing evaluation function F; 4. when the variation tendency of focusing back focusing evaluation function F value changed continuously, the readjustment focal length occurred until the image input signal corresponding to focusing evaluation function F maximum value, finishes the focusing process then; 5. get the image that promptly obtains sharpness the best of correct focusing corresponding to the image of the focusing evaluation function F maximum value of input signal.
2, the method for focusing automatically of a kind of digital image-forming based on the Contourlet conversion described in claim 1 is characterized in that described important directions comprises dominant direction, only adopts the significant coefficient of the determined subband of dominant direction to calculate focusing evaluation function.
3, a kind of digital image-forming based on the Contourlet conversion as claimed in claim 1 or 2 method of focusing automatically is characterized in that the Zone Full that two-dimentional input signal f is the digital micro imaging image or the combination or the down-sampled signal of regional area or regional area.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2008100922973A CN100543576C (en) | 2006-08-24 | 2006-08-24 | Based on the digital image-forming of the Contourlet conversion method of focusing automatically |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2008100922973A CN100543576C (en) | 2006-08-24 | 2006-08-24 | Based on the digital image-forming of the Contourlet conversion method of focusing automatically |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB2006100531186A Division CN100451808C (en) | 2006-08-24 | 2006-08-24 | Automatic focusing method of digital imaging based on Contourilct transform |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101266390A CN101266390A (en) | 2008-09-17 |
CN100543576C true CN100543576C (en) | 2009-09-23 |
Family
ID=39988893
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB2008100922973A Expired - Fee Related CN100543576C (en) | 2006-08-24 | 2006-08-24 | Based on the digital image-forming of the Contourlet conversion method of focusing automatically |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100543576C (en) |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110221953A1 (en) * | 2008-10-31 | 2011-09-15 | Stephen Pollard | method and digital imaging appliance for selecting a focus setting with a normalized figure-of-merit |
FR2949003B1 (en) * | 2009-08-10 | 2017-09-08 | Dxo Labs | SYSTEM AND METHOD FOR IMAGE CAPTURE WITH TWO OPERATING MODES |
US8027582B2 (en) * | 2009-12-21 | 2011-09-27 | Sony Corporation | Autofocus with confidence measure |
US9398205B2 (en) | 2010-09-01 | 2016-07-19 | Apple Inc. | Auto-focus control using image statistics data with coarse and fine auto-focus scores |
CN102915431A (en) * | 2012-09-12 | 2013-02-06 | 广东欧珀移动通信有限公司 | Method for automatically adjusting pictures according to terminal equipment picture contents |
WO2015139766A1 (en) * | 2014-03-21 | 2015-09-24 | Huawei Technologies Co., Ltd. | Imaging device and method for automatic focus in an imaging device as well as a corresponding computer program |
CN105118057A (en) * | 2015-08-18 | 2015-12-02 | 江南大学 | Image sharpness evaluation method based on quaternion wavelet transform amplitudes and phase positions |
CN108007675A (en) * | 2017-11-30 | 2018-05-08 | 福建福光股份有限公司 | The detection device and detection method of motorized zoom lens focusing curve |
CN108765346B (en) * | 2018-05-30 | 2021-01-08 | 北京图森智途科技有限公司 | Auxiliary focusing method and device and readable medium |
-
2006
- 2006-08-24 CN CNB2008100922973A patent/CN100543576C/en not_active Expired - Fee Related
Non-Patent Citations (4)
Title |
---|
基于Contourlet变换尺度间相关的图像去噪. 郁梅,易文娟,蒋刚毅.光电工程,第33卷第6期. 2006 |
基于Contourlet变换尺度间相关的图像去噪. 郁梅,易文娟,蒋刚毅.光电工程,第33卷第6期. 2006 * |
基于Contourlet变换的稳健性图像水印算法. 李海峰,宋巍巍,王树勋.通信学报,第27卷第4期. 2006 |
基于Contourlet变换的稳健性图像水印算法. 李海峰,宋巍巍,王树勋.通信学报,第27卷第4期. 2006 * |
Also Published As
Publication number | Publication date |
---|---|
CN101266390A (en) | 2008-09-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN100543576C (en) | Based on the digital image-forming of the Contourlet conversion method of focusing automatically | |
CN100451808C (en) | Automatic focusing method of digital imaging based on Contourilct transform | |
Hassen et al. | Image sharpness assessment based on local phase coherence | |
CN101520894B (en) | Method for extracting significant object based on region significance | |
CN102063713A (en) | Neighborhood normalized gradient and neighborhood standard deviation-based multi-focus image fusion method | |
CN103679672B (en) | Panorama image splicing method based on edge vertical distance matching | |
CN110097129A (en) | Remote sensing target detection method based on profile wave grouping feature pyramid convolution | |
CN101968883A (en) | Method for fusing multi-focus images based on wavelet transform and neighborhood characteristics | |
DE102007035884B4 (en) | Line noise suppression device, method and program | |
CN100573559C (en) | Self-adapting multi-dimension veins image segmenting method based on small echo and average drifting | |
CN103839265A (en) | SAR image registration method based on SIFT and normalized mutual information | |
CN102298773B (en) | Shape-adaptive non-local mean denoising method | |
DE102009036467A1 (en) | Pattern model positioning method in image processing, image processing apparatus, image processing program and computer-readable recording medium | |
CN105005565B (en) | Live soles spoor decorative pattern image search method | |
CN105719263A (en) | Visible light and infrared image fusion algorithm based on NSCT domain bottom layer visual features | |
CN102722877A (en) | Multi-focus image fusing method based on dual-channel PCNN (Pulse Coupled Neural Network) | |
CN110135438B (en) | Improved SURF algorithm based on gradient amplitude precomputation | |
CN103456006B (en) | A kind of oil depot object localization method of SAR image | |
DE60208255T2 (en) | Apparatus and method for image processing of finger and palm prints | |
CN108171656A (en) | Adaptive Global Dictionary remote sensing images ultra-resolution method based on rarefaction representation | |
CN112669249A (en) | Infrared and visible light image fusion method combining improved NSCT (non-subsampled Contourlet transform) transformation and deep learning | |
CN103914870A (en) | High-universality automatic hologram reestablishing method based on new focus evaluation function | |
CN110942423A (en) | Far-field super-resolution reconstruction method based on Fourier laminated imaging | |
CN1975762A (en) | Skin detecting method | |
CN111275680B (en) | SAR image change detection method based on Gabor convolution network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090923 Termination date: 20160824 |