CN118409729A - Road monitoring image denoising method and device - Google Patents
Road monitoring image denoising method and device Download PDFInfo
- Publication number
- CN118409729A CN118409729A CN202410882663.4A CN202410882663A CN118409729A CN 118409729 A CN118409729 A CN 118409729A CN 202410882663 A CN202410882663 A CN 202410882663A CN 118409729 A CN118409729 A CN 118409729A
- Authority
- CN
- China
- Prior art keywords
- wavelet
- function
- wavelet frame
- frame
- decomposition
- 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.)
- Granted
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
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method and a device for denoising a road monitoring image, which are characterized in that a wavelet frame filter bank is used for carrying out wavelet frame decomposition on the road monitoring image, the image is converted into a wavelet frame domain, and wavelet frames with different frequencies and time resolutions are obtained through multistage decomposition, namely the image is decomposed into a plurality of sub-bands; calculating to obtain a wavelet frame coefficient through the inner product of the sub-band and the wavelet frame function, and setting the absolute value of the wavelet frame coefficient to be 0, wherein the absolute value of the wavelet frame coefficient is lower than a preset threshold value; and finally, carrying out inverse transformation on the wavelet frame subjected to the threshold processing to obtain a repaired sub-band, and splicing all the sub-bands to obtain a denoised image. The method introduces a group of overcomplete wavelet basis functions on the basis of wavelet transformation, obtains a plurality of sub-bands through wavelet frame decomposition, can process noise areas more conveniently, and enhances the representation and reconstruction capability of images.
Description
Technical Field
The present invention relates to an image processing method and apparatus, and in particular, to an image denoising method and apparatus.
Background
In road monitoring, the image quality is of paramount importance, which directly affects the effectiveness and reliability of the monitoring system. The monitoring system with good image quality can provide clearer and more accurate images, and is convenient for better identifying and tracking the target, thereby improving the efficiency and the accuracy of the monitoring system. The monitoring system with poor image quality can cause problems of monitoring image blurring, pixel distortion, color distortion and the like, thereby influencing the reliability and accuracy of the monitoring system.
One common image quality problem is image noise, which is random brightness or color variation in an image, typically due to light, cameras, environmental factors, and the like. Noise can cause problems of distortion of a monitoring image, difficulty in identifying a target, false alarm and the like, so that reliability and accuracy of a monitoring system are reduced.
The basic idea of the image denoising method based on wavelet transform is to decompose an image containing noise into components with different frequencies through wavelet transform, and then process the components to achieve the purpose of removing the noise. The wavelet transformation is a mathematical transformation method, has the capability of multi-scale analysis, and can decompose an image into a group of wavelet basis functions with different scales and frequencies, so that the local characteristics and details of the image can be extracted. Wavelet transformation is generally decomposed by a binary decomposition-based method, and although the calculation amount is small, it is easy to cause the loss of low-frequency components, thereby affecting the restoration effect. And wavelet transforms decompose using orthogonal wavelet basis functions that have invariance and translational invariance but lack over-completeness, thereby limiting the representation and reconstruction capabilities of the image.
Disclosure of Invention
The invention aims to: aiming at the prior art, the method and the device for denoising the road monitoring image can effectively process a noise area and enhance the representation and reconstruction capability of the image.
The technical scheme is as follows: a method for denoising a road monitoring image, comprising: carrying out wavelet frame decomposition on the road monitoring image through a wavelet frame filter bank, converting the image into a wavelet frame domain, and obtaining wavelet frames with different frequencies and time resolutions through multistage decomposition, namely decomposing the image into a plurality of sub-bands; calculating to obtain a wavelet frame coefficient through the inner product of the sub-band and the wavelet frame function, and setting the absolute value of the wavelet frame coefficient to be 0, wherein the absolute value of the wavelet frame coefficient is lower than a preset threshold value; and finally, carrying out inverse transformation on the wavelet frame subjected to the threshold processing to obtain a repaired sub-band, and splicing all the sub-bands to obtain a denoised image.
Further, the wavelet frame filter bank comprises a low-pass filter h 0 (z) and a high-pass filter g 0 (z), and primary coefficients of the filters are respectively expressed as h 0 (n) and g 0 (n) and satisfy: wherein n represents the order of the filter; for the kth level wavelet frame decomposition, the corresponding filter coefficients are: Wherein h k (n) and g k (n) respectively represent coefficients of a low-pass filter and a high-pass filter used for decomposition of a kth-stage wavelet frame, m is a coefficient of a unit impulse response of the filter, i is an imaginary unit, and Z represents an integer domain;
in each stage of wavelet frame decomposition, the low-frequency signal is filtered by a low-pass filter to obtain an approximate signal of the next stage; filtering the high-frequency signal through a high-pass filter to obtain a detail signal of the next stage; wavelet frames with different frequencies and time resolutions are obtained through multistage decomposition.
Further, the Daubechies4 scale function and the Daubechies4 wavelet function are used as basic functions to construct the wavelet frame function; daubechies4 scale function is noted asThe Daubechies4 wavelet function is denoted as ψ (t), the functional expression is: wherein h k and g k represent the filter coefficients of the scale function and the wavelet function, respectively; n represents the order, k is the index of the wavelet coefficients;
and translating and expanding the scale function and the wavelet function: wherein j represents translation and expansion parameters and Z represents an integer domain; For the scale function after translation and expansion, ψ j,k (t) is the wavelet function for translation and expansion, and the function after translation and expansion operation is the wavelet frame function.
Further, the wavelet frame coefficient corresponding to the kth-level wavelet frame decomposition is calculated by the inner product of the sub-band and the wavelet frame function、The specific formula is as follows: where f (t) represents a subband function.
Further, the inner product calculation is realized through convolution operation, and specifically includes: discretizing wavelet frame functions, i.e. pairsAnd psi j,k (t) are discretized respectively to obtainAnd psi j,k (n), then calculating wavelet frame coefficients by a convolution mode, wherein a convolution formula is as follows: where L is the length of the wavelet frame function and f (n) represents the discretized subband function.
Further, the inverse transform is expressed as: where f' (t) represents the repaired subband.
A road monitoring image denoising device comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the road monitoring image denoising method.
Drawings
FIG. 1 is a functional image of a Daubechies4 scale function;
FIG. 2 is a functional image of a Daubechies4 wavelet function;
FIG. 3 is a functional image of a Daubechies4 wavelet frame function;
FIG. 4 is an image of a wavelet frame decomposition generation sub-band;
FIG. 5 is a schematic diagram of a partial wavelet frame coefficient thresholding process when thresholding 20;
FIG. 6 is an inverse transform of a repaired sub-band wavelet frame resulting in a repaired image;
Fig. 7 is a general flow chart of the present invention.
Detailed Description
The invention is further explained below with reference to the drawings.
As shown in fig. 7, a denoising method for road monitoring image specifically comprises the following steps:
Step 1: a set of basis functions is selected to construct a wavelet frame function.
In the invention, the Wavelet frame Function is obtained by a group of scale functions (Scaling functions) and Wavelet functions (Wavelet functions) through translation and expansion.
Specifically, the present embodiment constructs a wavelet frame function using a Daubechies4 scale function and a wavelet function as wavelet basis functions. The functional images of the Daubechies4 scale function and the wavelet function are shown in fig. 1 and 2, respectively, and the Daubechies4 scale function is recorded asThe Daubechies4 wavelet function is denoted as ψ (t), the function expression is as follows:
Wherein h k and g k represent filter coefficients of the scale function and the wavelet function, respectively; n denotes the order and k is the index of the wavelet coefficient.
And translating and expanding the scale function and the wavelet function, wherein the translation and expansion are as shown in the following formula:
wherein j represents translation and expansion parameters, and v2 j can ensure energy normalization of wavelet frame functions; z represents an integer domain; for the scale function after translation and expansion, ψ j,k (t) is the wavelet function for translation and expansion, the function after translation and expansion is the wavelet frame function, and the function image of the wavelet frame function is shown in fig. 3.
Step 2: and designing a wavelet frame filter bank, carrying out wavelet frame decomposition on the damaged road monitoring image, converting the image into a wavelet frame domain, and obtaining wavelet frames with different frequencies and time resolutions through multistage decomposition, namely decomposing the image into a plurality of sub-bands. The number of layers of the decomposition may be selected according to the size and complexity of the image.
The wavelet frame filter bank is a key part in the wavelet frame decomposition process, and consists of a low-pass filter h 0 (z) and a high-pass filter g 0 (z) and is used for converting an original image into a wavelet frame domain and decomposing the original image into wavelet frames with different frequencies.
In particular, the filter coefficients of the wavelet frame filter define the frequency response and phase response of the filter for low-pass and high-pass filtering of the image. For the low-pass filter H 0 (z) and the high-pass filter G 0 (z), their frequency response functions are H 0 (z) and G 0 (z), respectively, and their filter coefficients are expressed as:
where n is an integer representing the order of the filter. For the kth level wavelet frame decomposition, the corresponding filter coefficients are shown as follows:
Where h k (n) and g k (n) represent coefficients of a low-pass filter and a high-pass filter used for the kth-stage wavelet frame decomposition, respectively, m is a coefficient of a unit impulse response of the filter, i is an imaginary unit, and Z represents an integer domain.
In each stage of wavelet frame decomposition, the low-frequency signal is filtered by a low-pass filter to obtain an approximate signal of the next stage; filtering the high-frequency signal through a high-pass filter to obtain a detail signal of the next stage; wavelet frames of different frequencies and time resolutions are obtained by multi-stage decomposition, i.e. the image is decomposed into a plurality of sub-bands by a filter bank.
In this embodiment, 3 layers of wavelet frame decomposition is selected, as shown in fig. 4, a noise picture is input, and each subband is obtained after three levels of wavelet frame decomposition, where the horizontal and vertical coordinates in the figure represent pixel coordinates in the wavelet domain.
Step 3: wavelet frame coefficients are calculated.
The wavelet frame coefficient is a decomposition coefficient used for representing a signal on the basis of a wavelet frame function, and is obtained by calculating the inner product of a subband and the wavelet frame function, and the specific formula is as follows:
Where f (t) represents a subband function.
The calculation process is realized efficiently through convolution operation, and specifically comprises the following steps: discretizing wavelet frame functions, i.e. pairsAnd psi j,k (t) are discretized respectively to obtainAnd phi j,k (n), and then calculating wavelet frame coefficients by means of convolution. Assuming that the length of the wavelet frame function is L, the convolution formula is as follows:
Where f (n) represents a discretized subband function.
Step 4: the wavelet frame coefficients are analyzed and thresholded.
The image noise repairing process is to perform threshold processing on wavelet frame coefficients so as to reduce the influence of noise and other interference components. Specifically, the root mean square value of the signal, that is, the square root of the mean value of the square of the signal is calculated according to the distribution of the wavelet coefficients, and it is common to set the threshold τ to a certain multiple of the root mean square value, and this embodiment selects 2 times, and then sets the value of the absolute value of the wavelet frame coefficients smaller than the threshold to 0, so that smaller noise components can be removed and larger signal components can be retained. The thresholding is expressed by a mathematical formula as follows:
Wherein, AndRepresenting the thresholded wavelet coefficients.
As shown in fig. 5, which is a processing flow of wavelet coefficients at a threshold of 20, it can be seen that the value of the absolute value of the coefficient of 20 or less is set to 0, while the other coefficients remain unchanged.
Step 5: and carrying out inverse transformation on the wavelet frame subjected to the threshold processing to obtain a repaired sub-band, and splicing all the sub-bands to obtain a repaired image. Wherein, the inverse transformation formula of the wavelet frame transformation is as follows:
Where f' (t) represents the repaired subband, the first summation term represents the contribution of the scale factor, and the second summation term represents the contribution of the wavelet factor.
Fig. 6 shows a restored image obtained by inverse wavelet frame transformation of the restored subband in this embodiment, where the horizontal and vertical coordinates represent pixel coordinates in the wavelet domain.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.
Claims (7)
1. A method for denoising a road monitoring image, comprising: carrying out wavelet frame decomposition on the road monitoring image through a wavelet frame filter bank, converting the image into a wavelet frame domain, and obtaining wavelet frames with different frequencies and time resolutions through multistage decomposition, namely decomposing the image into a plurality of sub-bands; calculating to obtain a wavelet frame coefficient through the inner product of the sub-band and the wavelet frame function, and setting the absolute value of the wavelet frame coefficient to be 0, wherein the absolute value of the wavelet frame coefficient is lower than a preset threshold value; and finally, carrying out inverse transformation on the wavelet frame subjected to the threshold processing to obtain a repaired sub-band, and splicing all the sub-bands to obtain a denoised image.
2. The method for denoising road monitoring image according to claim 1, wherein the wavelet frame filter bank comprises a low-pass filter h 0 (z) and a high-pass filter g 0 (z), and the primary coefficients of the filters are respectively denoted as h 0 (n) and g 0 (n) and satisfy: wherein n represents the order of the filter; for the kth level wavelet frame decomposition, the corresponding filter coefficients are: Wherein h k (n) and g k (n) respectively represent coefficients of a low-pass filter and a high-pass filter used for decomposition of a kth-stage wavelet frame, m is a coefficient of a unit impulse response of the filter, i is an imaginary unit, and Z represents an integer domain;
in each stage of wavelet frame decomposition, the low-frequency signal is filtered by a low-pass filter to obtain an approximate signal of the next stage; filtering the high-frequency signal through a high-pass filter to obtain a detail signal of the next stage; wavelet frames with different frequencies and time resolutions are obtained through multistage decomposition.
3. The method for denoising a road monitoring image according to claim 1 or 2, wherein the wavelet frame function is constructed using a Daubechies4 scale function and a Daubechies4 wavelet function as basis functions; daubechies4 scale function is noted asThe Daubechies4 wavelet function is denoted as ψ (t), the functional expression is: wherein h k and g k represent the filter coefficients of the scale function and the wavelet function, respectively; n represents the order, k is the index of the wavelet coefficients;
and translating and expanding the scale function and the wavelet function: wherein j represents translation and expansion parameters and Z represents an integer domain; For the scale function after translation and expansion, ψ j,k (t) is the wavelet function for translation and expansion, and the function after translation and expansion operation is the wavelet frame function.
4. A method for denoising a road monitoring image according to claim 3, wherein the wavelet frame coefficients corresponding to the kth level wavelet frame decomposition are calculated by the inner product of the subband and the wavelet frame function、The specific formula is as follows: where f (t) represents a subband function.
5. The method for denoising a road monitoring image according to claim 4, wherein the inner product calculation is realized by convolution operation, specifically comprising: discretizing wavelet frame functions, i.e. pairsAnd psi j,k (t) are discretized respectively to obtainAnd psi j,k (n), then calculating wavelet frame coefficients by a convolution mode, wherein a convolution formula is as follows: where L is the length of the wavelet frame function and f (n) represents the discretized subband function.
6. The road monitoring image denoising method according to claim 5, wherein the inverse transformation is expressed as: where f' (t) represents the repaired subband.
7. A road monitoring image denoising apparatus comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor executing the program to implement the road monitoring image denoising method as claimed in any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410882663.4A CN118409729B (en) | 2024-07-03 | 2024-07-03 | Road monitoring image denoising method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410882663.4A CN118409729B (en) | 2024-07-03 | 2024-07-03 | Road monitoring image denoising method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118409729A true CN118409729A (en) | 2024-07-30 |
| CN118409729B CN118409729B (en) | 2024-08-23 |
Family
ID=92003540
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410882663.4A Active CN118409729B (en) | 2024-07-03 | 2024-07-03 | Road monitoring image denoising method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118409729B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6141452A (en) * | 1996-05-13 | 2000-10-31 | Fujitsu Limited | Apparatus for compressing and restoring image data using wavelet transform |
| US6826513B1 (en) * | 2002-11-29 | 2004-11-30 | Council Of Scientific & Industrial Research | Method and apparatus for online identification of safe operation and advance detection of unsafe operation of a system or process |
| US7239424B1 (en) * | 2000-09-08 | 2007-07-03 | Ricoh Co., Ltd. | Wavelet-based image processing path |
| CN105894477A (en) * | 2016-06-03 | 2016-08-24 | 深圳市樊溪电子有限公司 | Astronomical image noise removal method |
| CN115272111A (en) * | 2022-07-22 | 2022-11-01 | 中国人民解放军战略支援部队信息工程大学 | A method and system for noise reduction of sonar image based on shear wave transform based on Meyer window function |
-
2024
- 2024-07-03 CN CN202410882663.4A patent/CN118409729B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6141452A (en) * | 1996-05-13 | 2000-10-31 | Fujitsu Limited | Apparatus for compressing and restoring image data using wavelet transform |
| US7239424B1 (en) * | 2000-09-08 | 2007-07-03 | Ricoh Co., Ltd. | Wavelet-based image processing path |
| US6826513B1 (en) * | 2002-11-29 | 2004-11-30 | Council Of Scientific & Industrial Research | Method and apparatus for online identification of safe operation and advance detection of unsafe operation of a system or process |
| CN105894477A (en) * | 2016-06-03 | 2016-08-24 | 深圳市樊溪电子有限公司 | Astronomical image noise removal method |
| CN115272111A (en) * | 2022-07-22 | 2022-11-01 | 中国人民解放军战略支援部队信息工程大学 | A method and system for noise reduction of sonar image based on shear wave transform based on Meyer window function |
Non-Patent Citations (3)
| Title |
|---|
| LI ZHANG: "Research on LIF Image Denoising Based on Wavelet-Domain Multiscale HMT Model", 2006 6TH WORLD CONGRESS ON INTELLIGENT CONTROL AND AUTOMATION, 23 October 2006 (2006-10-23) * |
| 沈腾飞: "紧小波框架对高维Sobolev空间的刻画与图像去噪研究", CNKI优秀硕士学位论文全文库, 25 February 2023 (2023-02-25) * |
| 王再奎, 严卫, 马宁, 王庆东: "基于复对称Daubechies小波分析的SAR图像去噪", 解放军理工大学学报(自然科学版), no. 06, 20 January 2003 (2003-01-20) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118409729B (en) | 2024-08-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7260272B2 (en) | Method and apparatus for noise reduction using discrete wavelet transform | |
| CN101390108B (en) | System and method for performing wavelet-based texture feature extraction and classification | |
| Zhang et al. | Joint image denoising using adaptive principal component analysis and self-similarity | |
| CN102117482B (en) | Non-local mean image denoising method combined with structural information | |
| KR20190031295A (en) | Background suppression method and detection device during display panel automatic optical detection | |
| CN112991199B (en) | Image high-low frequency decomposition noise removal method based on residual dense network | |
| CN112037151A (en) | Image denoising method based on wavelet analysis | |
| CN117911267A (en) | Back-scattered X-ray image noise reduction method based on wavelet transformation | |
| CN113436078B (en) | Self-adaptive image super-resolution reconstruction method and device | |
| CN118409729B (en) | Road monitoring image denoising method and device | |
| Panigrahi et al. | Joint bilateral filter for signal recovery from phase preserved curvelet coefficients for image denoising | |
| CN118014885A (en) | Method and device for eliminating background noise and storage medium | |
| Sakiyama et al. | Edge-aware image graph expansion methods for oversampled graph Laplacian matrix | |
| Bhonsle et al. | Medical image de-noising using combined bayes shrink and total variation techniques | |
| CN106780398A (en) | A kind of image de-noising method based on noise prediction | |
| CN111402166A (en) | Image denoising method and device, service terminal and computer readable storage medium | |
| Verma et al. | Analysis of Hybrid Wavelet Transformed Images to Improve Image Pixel Quality | |
| Kumar et al. | A mixture of Noise Image Denoising using Sevenlets Wavelet Techniques | |
| Foucher et al. | Unsupervised multiscale speckle filtering | |
| Talbi et al. | A hybrid technique of image denoising using the curvelet transform based denoising method and two-stage image denoising by PCA with local pixel grouping | |
| Wu | Wavelet Analysis for Image Denoising: A Multiscale Approach for Enhancing Visual Clarity | |
| Saood et al. | An efficient image denoising approach using FPGA type of PYNQ-Z2 | |
| CN113962894B (en) | A screen image processing method | |
| Shree et al. | A new fusion-based agricultural synthetic aperture radar image despeckling by using anisotropic diffusion and discrete wavelet transform methods | |
| Jabbar et al. | Transformation based image de-noising |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |