CN1892698A - Remote-sensing picture interpolation method based on small wave fractal - Google Patents
Remote-sensing picture interpolation method based on small wave fractal Download PDFInfo
- Publication number
- CN1892698A CN1892698A CN 200610081450 CN200610081450A CN1892698A CN 1892698 A CN1892698 A CN 1892698A CN 200610081450 CN200610081450 CN 200610081450 CN 200610081450 A CN200610081450 A CN 200610081450A CN 1892698 A CN1892698 A CN 1892698A
- Authority
- CN
- China
- Prior art keywords
- fractal
- image
- interpolation
- wavelet
- remote
- 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
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
A wavelet fractal based on remotely sensed image interplation method adopts wavelet analysing and fractal technology adjoint to realize remotely sensed image interplation. It contains firstly making wavelet conversion to remotely sensed image to obtain remotely sensed image one low-frequency component and three high frequency component, dividing each high frequency component into mutually exclusive subimage block, utilizing image fractal character to calculate each subimage block fractal parameter and making self adaptive fractal interplation to every subimage block, then merging high frequency component subimage block and original image after fractal interplation, finally making inverse wavelet conversion to merged image to obtain high resolution remotely sensed image processed by wavelet fractal interplation.
Description
Technical field
The present invention proposes a kind of remote-sensing picture interpolation method based on small wave fractal in image processing field.Adopt the wavelet analysis realization remote-sensing picture interpolation that combines with fractal technology.The small wave fractal interpolation is that remote sensing images are done similarity transformation with fractal on the basis of wavelet transformation, and then obtains the interpolation image higher than original image spatial resolution by inverse wavelet transform.The smoothing effect that traditional interpolation method (as spline interpolation, bilinear interpolation) causes be can effectively overcome, remote sensing images abundant texture and target edges kept.
Adopt the wavelet analysis realization remote-sensing picture interpolation that combines with fractal technology.At first remote sensing images are carried out wavelet transformation, obtain a low frequency component and three high fdrequency components of remote sensing images.Each high fdrequency component is divided into mutually disjoint subimage block, utilizes the fractal property of image, calculate the fractal parameter of each subimage block and each subimage block is carried out adaptive fractal interpolation.Merge through high fdrequency component subimage block and original image behind the fractal interpolation then.Image after will merging is at last made inverse wavelet transform, then obtains through the high-resolution remote sensing image after the small wave fractal interpolation processing.
Background technology
Early stage remote-sensing picture interpolation mainly uses arest neighbors interpolation, linear interpolation, bilinear interpolation or spline interpolation scheduling algorithm, and these interpolation algorithms have been released the hardware of algorithm and realized because complexity is low, can obtain satisfied result under the not high situation of accuracy requirement.And the remotely sensed image zone is big, cause remote sensing images resolution low, various sea situations, landform, landforms form complex texture (as wave of the sea, forest etc.) and target edges (as seashore, artificiality etc.), adopt above-mentioned interpolation algorithm to improve the remote sensing images spatial resolution, smoothing effect and pseudo-shadow have been produced at texture and edge, seriously undermine the artificial and machine decipher of remote sensing images, reduced the application efficiency of remote sensing images.
For improving the performance of remote-sensing picture interpolation algorithm, can carry out texture and target edges detection to the input remote sensing images, when carrying out interpolation, can carry out the emphasis processing like this to remote sensing images texture and target edges, and then suppress the smoothing effect that remote-sensing picture interpolation produced, improve the remote sensing images application level.(sky) frequency analysis theory when wavelet theory is rising in recent years a kind of brand-new.Mallat at first is used for wavelet transformation the decomposition and reconstruction of computer picture, has proposed the notion of multiresolution analysis.The iamge description of this multiresolution is called the wavelet decomposition of image.Can decompose into some to picture signal with wavelet transformation and have different spatial resolutions, the picture content of frequency characteristic and directivity characteristics, and have similarity between the equidirectional component, thereby wavelet analysis can be applicable to remote-sensing picture interpolation, improves image spatial resolution.
Wavelet function is a tight function of local non-zero, and (Wavelet Transform, WT) the wavelet coefficient reality after can be regarded the result who handles through high frequency filter as to the remote sensing images wavelet transformation.According to multiresolution analysis thought, wavelet transformation is equivalent to a pair of bank of filters of being made up of high pass and low-pass filter, and this is convenient to realize the quick wavelet decomposition and the reconstruct of image.In conjunction with the character of Hi-pass filter as can be known, remote sensing images are through behind the wavelet transformation, will produce local extremum at the image part wavelet coefficient of undergoing mutation.According to this local characteristics of wavelet coefficient, can detect the texture and the target edges of remote sensing images.
Mandelbrot has proposed fractal geometry, describes spontaneous phenomenon complicated and changeable with this notion of fractal dimension, points out that the similar in some way shape of ingredient and integral body is fractal.A.P.Pentland proves that the image that the surface was mapped to of most of natural scenes satisfies fractal Blang's random field models (FBR), and with it as natural scene analyzing image texture and comprehensive a kind of model.A fractal essential characteristic be can represented object fine structure, if can extract the fractal parameter of image, utilize these parameters just can reach the purpose that under arbitrary resolution, generates photorealism, also just realized the level and smooth amplification effect of image.
Have the characteristic of random fractal Blang field (FBR) model according to image, the mark Blang random field of image is defined as:
If X, Δ X ∈ R
2, 0≤H≤1, F (y) is that average is zero gaussian random function, P
r() expression probability measure, ‖ ‖ represents norm, if random field B
H(X) satisfy:
B then
H(X) be mark Blang random field (FBR).H is relevant with the roughness height on fractal pattern surface, and the fractal dimension D that can be got imaging surface by the H parameter is D=D
T+ 1-H, wherein D
TTopological dimension for imaging surface.
B
H(X) have character:
E|B
H(X+ΔX)-B
H(X)|
2＝E|B
H(X+1)-B
H(X)|
2||ΔX||
2H (2)
The pass that can draw H and σ according to (2) formula is:
log?E|B
H(X+ΔX)-B
H(X)|
2-2H?log|ΔX|＝logσ
2 (3)
In the formula, normal distribution standard variance σ
2=E|B
H(X+1)-B
H(X) |
2
Fractal interpolation comes down to a kind of process of recurrence neutral displacement.Its interpolation method is:
A) i, when j is odd number, point (i, j) gray-scale value B
HDetermine.
B) i, when j is even number:
C) work as i, when being even number one of among the j:
In the fractal interpolation processing procedure of image, main parameter has fractal characteristic parameter H, normal distribution standard deviation sigma and Gaussian stochastic variable G.G obeys N (0,1) and distributes, || Δ X|| is a sample separation.G produces by the pseudo-random function simulation of computing machine.And fractal characteristic parameter H and normal distribution standard deviation sigma are calculated by FBR field model (3) formula of image and by the characteristics of image least square method.
But in fact image is not desirable fractal fully, neither the FRACTAL DIMENSION number average keeps constant under any yardstick, a range scale is arranged.Therefore, need to select the scope of Δ x, so that make in this scope, H is a constant. (3) formula log E|B in a segment limit
H(X+ Δ X)-B
H(X) |
2X| is linear with the log| Δ, finds out linear bound | Δ X|
MinWith | Δ X|
Max, it is the scope of constant that this section is maintenance H, can calculate H and σ with least square method in this bound scope.
Remotely sensed image is observation of nature scenery under large scale mainly, and under the large scale prerequisite, natural scene has self-similarity or local self-similarity, and fractal (Fractal) has natural advantage aspect description self-similarity or the local self-similarity image.Though fractally can effectively reduce smoothing effect, keep the texture information of original image, through repeatedly behind the fractal interpolation tangible smoothing effect being arranged still, can not keep texture information.Remote sensing images are done similarity transformation with fractal on the basis of wavelet transformation, and then obtain the interpolation image higher than original image resolution by inverse transformation.This new small wave fractal interpolation method that wavelet analysis is combined with fractal technology.The smoothing effect that traditional interpolation method (as spline interpolation, bilinear interpolation) causes be can effectively overcome, remote sensing images abundant texture and target edges kept.
Summary of the invention
In order to solve smoothing effect and the pseudo-shadow problem that produces in traditional remote-sensing picture interpolation algorithm, the present invention combines wavelet analysis with fractal technology, has proposed a kind of implementation method of the remote-sensing picture interpolation based on small wave fractal.The small wave fractal interpolation method is: at first remote sensing images are carried out wavelet transformation, obtain a low frequency component and three high fdrequency components of remote sensing images.Each high fdrequency component is divided into mutually disjoint image subblock, utilizes the fractal property of image, calculate the fractal parameter of each image subblock and each subimage block is carried out adaptive fractal interpolation.Merge through high fdrequency component subimage block and original image behind the fractal interpolation.Image after will merging is at last made inverse wavelet transform, then obtains through the high-resolution remote sensing image after the small wave fractal interpolation processing.
The technical solution adopted for the present invention to solve the technical problems is: image is divided into a low frequency component image LL to original image through behind the wavelet transformation
1With horizontal direction LH
1, vertical direction HL
1And diagonal HH
1Three high fdrequency component images.LH
1The texture and the marginal texture of expression original image horizontal direction.HL
1The texture and the marginal texture of expression original image side vertical direction.HH
1The relevant information of expression original image diagonal.Because abundant texture and the details of image is included in HFS,, can make interpolation image keep the texture of original image like this so adopt the method for fractal interpolation to obtain three high fdrequency components of original image.Different with the small echo bilinear interpolation is that when calculating the gray scale of interpolation point, the small wave fractal interpolation has utilized the fractal property of image that the gray scale of interpolation point has been carried out the suitable processing of self-adaptation adjustment.The method of small wave fractal interpolation is: each high fdrequency component is divided into mutually disjoint image subblock, calculate the fractal parameter H and the σ of each image subblock earlier, utilize foregoing interpolation formula (4) and (5) to carry out fractal interpolation, near interpolation point H value hour, the key diagram picture rises and falls near interpolation point greatly, needs interpolation point is done big adjustment; Otherwise when near the H value the interpolation point was big, the key diagram picture rose and fell little near interpolation point, only needed the little adjustment of interpolation point.This interpolation process can be performed until and satisfy till the desired resolution.Merge through high fdrequency component subimage block and original image (replacing low-frequency image with original image) behind the fractal interpolation then, the image after will merging is at last made inverse wavelet transform, then obtains through the high-resolution remote sensing image after the small wave fractal interpolation processing.
The invention has the beneficial effects as follows, by remote sensing images are carried out the small wave fractal interpolation processing, can improve the remote sensing images spatial resolution, obtain more accurate artificial and machine decipher result, improve remote sensing images application level, particularly remote sensing images robotization processing horizontal.
Description of drawings
Fig. 1 is the fractal interpolation system model
In the drawings: 1. image segmentating device, 2. fractal parameter counter, 3. the bilinear interpolation device 4. improves totalizer, 5. image combiner, x is the input remote sensing images, y is the remote sensing images after handling through fractal interpolation.
Fig. 2 is the small wave fractal interpolation model
In the drawings: 11. wavelet transformer, 12. fractal interpolation systems, 13. inverse wavelet, x is through the remote sensing images after the small wave fractal interpolation processing for input remote sensing images, z.
The detailed structure of fractal interpolation device 12 as shown in Figure 1 among Fig. 2.
Embodiment
Based on the specific embodiments of the remote-sensing picture interpolation method of small wave fractal as shown in Figure 1 and Figure 2, specific embodiments is:
At first the input remote sensing images are carried out the one-level wavelet transformation.Remote sensing images x 11 carries out wavelet transformation in Fig. 2, obtain a low frequency component LL
1With three high fdrequency component LH
1, HL
1And HH
1The wavelet function of selecting is a separable function, i.e. the 2-d wavelet function.The 2-d wavelet function can be decomposed into the product of two one dimension wavelet functions, thereby the wavelet transformation of remote sensing images is equal to the row, column of image is made one-dimensional wavelet transform respectively (when the image reconstruction, adopt corresponding order to carry out inverse wavelet transform, promptly row, the row to changing image carries out the one dimension inverse wavelet transform).
The fractal interpolation of remote sensing images high fdrequency component 12 carries out (being to carry out among Fig. 1) behind the wavelet transformation in Fig. 2.At first each high fdrequency component image is divided into mutually disjoint subimage block x in the image segmentating device 1 in Fig. 1
1, x
2..., x
n, again with each subimage block x
i(i=1,2 ..., n) fractal parameter counter 2 and bilinear interpolation device 3 among input Fig. 1, the output of fractal parameter counter is through the fractal parameter f of the subimage block after calculating
i(i=1,2 ..., n), the subimage block x after the output of bilinear interpolation device is handled through bilinear interpolation
i' (i=1,2 ..., n).And then, obtain the subimage block x that handles through fractal interpolation with improving totalizer 4 among fi and xi ' input Fig. 1
i" (i=1,2 ..., n).Fractal interpolation subimage block x
1", x
2" ..., x
n" import again that image combiner 5 obtains fractal interpolation image y among Fig. 1.Wherein, the improvement totalizer 4 in the fractal interpolation device is compared with common totalizer, has increased the Intelligence Selection function, promptly according to the position of interpolation point, and the fractal side-play amount that automatic selection need add.This side-play amount is calculated according to fractal parameter, the self-similarity of principal security input picture or local self-similarity.
At last, with former remote sensing images with through three high fdrequency component LH after the fractal interpolation processing
1', HL
1' and HH
1' merging, inverse wavelet 13 among input Fig. 2 is carried out inverse wavelet transform, is output as through the high-resolution remote sensing image z after the small wave fractal interpolation processing.
Claims (3)
1. based on the remote-sensing picture interpolation method of small wave fractal, it is characterized in that split image piece in the high-frequency sub-band of wavelet transformed domain, has similarity according to the high fdrequency component on equidirectional, on the different scale, can detect the object video of high-frequency sub-band at different levels by the set membership of equidirectional locus of going up the high-frequency sub-band object video and wavelet coefficient, realize image segmentation.
2. the remote-sensing picture interpolation method based on small wave fractal according to claim 1, its feature also is: each high fdrequency component is divided into mutually disjoint subimage block, utilize the fractal property of image, calculate the fractal parameter of each subimage block and each subimage block is carried out adaptive fractal interpolation.
3. the remote-sensing picture interpolation method based on small wave fractal according to claim 1 and 2, its feature also is: merge through high fdrequency component subimage block and original image behind the fractal interpolation, and the image after will merging makes inverse wavelet transform, then obtains through the high-resolution remote sensing image after the small wave fractal interpolation processing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200610081450 CN1892698A (en) | 2006-05-19 | 2006-05-19 | Remote-sensing picture interpolation method based on small wave fractal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200610081450 CN1892698A (en) | 2006-05-19 | 2006-05-19 | Remote-sensing picture interpolation method based on small wave fractal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1892698A true CN1892698A (en) | 2007-01-10 |
Family
ID=37597554
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200610081450 Pending CN1892698A (en) | 2006-05-19 | 2006-05-19 | Remote-sensing picture interpolation method based on small wave fractal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1892698A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101201900B (en) * | 2007-11-06 | 2010-09-15 | 重庆大学 | Method for regulating human face image illumination based on multilevel wavelet disintegrating and spline interpolation |
| CN101835045A (en) * | 2010-05-05 | 2010-09-15 | 哈尔滨工业大学 | Hi-fidelity remote sensing image compression and resolution ratio enhancement joint treatment method |
| CN102047287B (en) * | 2008-06-17 | 2013-03-13 | 株式会社Ntt都科摩 | Image/video quality enhancement and super-resolution using sparse transformations |
| CN104574346A (en) * | 2013-10-23 | 2015-04-29 | 核工业北京地质研究院 | Optical remote sensing image decomposition algorithm |
| CN106600542A (en) * | 2016-10-31 | 2017-04-26 | 北京空间机电研究所 | Spaceflight optical remote sensing high-density quantization information processing method |
| CN113191989A (en) * | 2021-05-25 | 2021-07-30 | 宁波大学 | Water line remote sensing extraction method for muddy coast under near-shore complex environment |
-
2006
- 2006-05-19 CN CN 200610081450 patent/CN1892698A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101201900B (en) * | 2007-11-06 | 2010-09-15 | 重庆大学 | Method for regulating human face image illumination based on multilevel wavelet disintegrating and spline interpolation |
| CN102047287B (en) * | 2008-06-17 | 2013-03-13 | 株式会社Ntt都科摩 | Image/video quality enhancement and super-resolution using sparse transformations |
| CN101835045A (en) * | 2010-05-05 | 2010-09-15 | 哈尔滨工业大学 | Hi-fidelity remote sensing image compression and resolution ratio enhancement joint treatment method |
| CN104574346A (en) * | 2013-10-23 | 2015-04-29 | 核工业北京地质研究院 | Optical remote sensing image decomposition algorithm |
| CN106600542A (en) * | 2016-10-31 | 2017-04-26 | 北京空间机电研究所 | Spaceflight optical remote sensing high-density quantization information processing method |
| CN106600542B (en) * | 2016-10-31 | 2020-04-10 | 北京空间机电研究所 | Aerospace optical remote sensing high-density quantization information processing method |
| CN113191989A (en) * | 2021-05-25 | 2021-07-30 | 宁波大学 | Water line remote sensing extraction method for muddy coast under near-shore complex environment |
| CN113191989B (en) * | 2021-05-25 | 2021-11-19 | 宁波大学 | Water line remote sensing extraction method for muddy coast under near-shore complex environment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sandić-Stanković et al. | DIBR synthesized image quality assessment based on morphological wavelets | |
| CN101478693B (en) | Method for evaluating star-loaded optical remote sensing image compression quality | |
| CN105574534A (en) | Significant object detection method based on sparse subspace clustering and low-order expression | |
| CN105931257B (en) | SAR image method for evaluating quality based on textural characteristics and structural similarity | |
| CN105335972B (en) | Knitted fabric defect detection method based on small echo contourlet transform and vision significance | |
| CN1892698A (en) | Remote-sensing picture interpolation method based on small wave fractal | |
| CN102572499B (en) | Based on the non-reference picture quality appraisement method of wavelet transformation multi-resolution prediction | |
| CN110532615B (en) | Progressive decomposition method for shallow sea complex terrain | |
| CN101847257A (en) | Image denoising method based on non-local means and multi-level directional images | |
| CN104834931A (en) | Improved SIFT algorithm based on wavelet transformation | |
| CN103065320A (en) | Synthetic aperture radar (SAR) image change detection method based on constant false alarm threshold value | |
| Zhao et al. | An approach to compute fractal dimension of color images | |
| CN102819829A (en) | Rebuilding algorithm for super-resolution remote sensing image based on fractal theory | |
| CN105809650A (en) | Bidirectional iteration optimization based image integrating method | |
| CN113096015B (en) | Image super-resolution reconstruction method based on progressive perception and ultra-lightweight network | |
| CN102142130B (en) | Watermark embedding method and device based on wavelet-domain enhanced image masks | |
| CN108090914B (en) | Color image segmentation method based on statistical modeling and pixel classification | |
| CN105469358A (en) | Image processing method | |
| CN106096650B (en) | Based on the SAR image classification method for shrinking self-encoding encoder | |
| CN113205564A (en) | SAR intelligent target edge reconstruction method | |
| CN109447952B (en) | Semi-reference image quality evaluation method based on Gabor differential box weighting dimension | |
| CN102298768B (en) | High-resolution image reconstruction method based on sparse samples | |
| CN107767342B (en) | Wavelet transform super-resolution image reconstruction method based on integral adjustment model | |
| Wang | Single image super-resolution with u-net generative adversarial networks | |
| CN113077488B (en) | River surface flow velocity detection method and device |
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 |