CN103886551A - Image filtering processor - Google Patents
Image filtering processor Download PDFInfo
- Publication number
- CN103886551A CN103886551A CN201310606022.8A CN201310606022A CN103886551A CN 103886551 A CN103886551 A CN 103886551A CN 201310606022 A CN201310606022 A CN 201310606022A CN 103886551 A CN103886551 A CN 103886551A
- Authority
- CN
- China
- Prior art keywords
- window
- value
- image
- image filtering
- new
- 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
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses an image filtering processor, and aims at enhancing processing speed of median filtering under the situation of fully considering and utilizing the related information of adjacent pixels and characteristics of window movement. The image filtering processor comprises four steps of a window movement step, a condition judgment step, a numerical value ordering step and a new value output step. The realization method of the aforementioned steps is that: after the window movement step is performed, the condition judgment step is performed on data in a window, and an original mid-value is outputted if conditions are established, or the numerical value ordering step is performed. After that the new value output step is performed, and then a new round of cycle is started.
Description
Technical field
The present invention relates to image filtering field, more specifically a kind of by the implementation of existing medium filtering being improved to the image filtering processor of rear formation, to improve the speed of image filtering.
Background technology
At present, in automatic visual detecting system because the collection of image is to carry out on the travelling belt of linear uniform motion substantially, based on very many-sided reason, the image collecting in monitoring site generally all can not react the true content of scene completely, produced the problems such as image fault, this phenomenon is referred to as image degradation.The degeneration producing in image acquisition process is often called as fuzzy, and it is to the restricted effect of the spectrum width of target.The degeneration producing in image capture process is often called as noise, and it can derive from measuring error, counting error etc.
For instance, lens chromatic aberration (aberration) is a kind of inherent shortcoming of optical system, and it has limited the sharpness of definition of the image being obtained by shot by camera.Backward, although there is a desirable optical system, the acutance of image also can be because electromagnetic diffraction is restricted.The degeneration of these types is intrinsic properties of given optical system.In addition, due to focus on that inaccurate (out of focus) cause image blurring be a kind of common due to transient cause, as adjust mistake and the image degradation that causes, the acutance that it also can limited images.So be the main cause of image degradation image blurring, the present invention analyzes emphasis for this situation.
At present, the domestic research to still image still has correlation technique need to capture or perfect, comprises Denoising Problems, and image is cut apart and the technical matterss such as edge extracting.Still image processing is the basis of dynamic image processing.
Noise is one of modal degeneration factor, is also the content of primary study during image recovers.Concerning signal, noise is a kind of external disturbance.But noise itself is also a kind of signal, and only it has carried the information of noise source.The gray scale of noise itself can be regarded stochastic variable as, and its distribution usable probability density function (PDF) is described, and is several important noise probability density functions below.
L) Gaussian noise
The probability density function of a Gaussian random variable z can be expressed as:
Wherein z represents gray scale, and μ is the average of z, and σ is the standard deviation of z.
2) even noise
Evenly the probability density function of noise can be expressed as:
Evenly the average of noise and variance are respectively:
3) pulse (spiced salt) noise
The probability density function of impulsive noise can be expressed as:
In image is processed, as the typical nonlinear filtering algorithm of one, medium filtering application is very extensively in removing noise.Medium filtering, compared with linear filtering, has the valuable property of Retain edge information.As a kind of special order statistic filtering, it has stability, fixed point and keeps the feature of local monotonicity.Medium filtering is a kind of nonlinear signal processing technology of being invented by Tukey, in early days as signal processing, be used to very soon afterwards two-dimensional digital image level and smooth in.Adopt common linear low-pass filtering method, in denoising, often make original clear profile in image fog, and people's vision is very sensitive to the edge of image.The outstanding advantages of medium filtering is in keeping the original clear profile of image, and the noise filtering of image is fallen, and especially salt-pepper noise and impulsive noise is had to better effect.But the data sorting in common median filtering algorithm is time-consuming more, need to carry out a large amount of data and relatively work, be unfavorable for the online fast processing of image.
Summary of the invention
For addressing the above problem, the invention discloses a kind of image filtering processor, in the case of taking into full account and utilize the relevant information of neighbor and feature that window moves, object is to improve the processing speed of medium filtering.
The present invention takes following technical scheme to realize: a kind of image filtering processor, comprises that window moves, condition judgment, numerical ordering and new four steps of value of output.The implementation of above-mentioned steps is to the data executive condition determining step in window after carrying out window to move step, if condition is set up, export former intermediate value, otherwise carry out numerical value ordered steps, carry out afterwards the new value of output step, then enter the circulation of a new round.
Realization of the present invention also comprises following technical scheme:
It is the implementation that the window of former medium filtering moves that above-mentioned window moves step.
Whether the condition of above-mentioned condition judgment step for meeting i=l & & j=m & & k=n, if set up, utilized the limited and neighbor of quantification number of greyscale levels to have very large correlativity feature and exported former intermediate value.
The implementation of above-mentioned numerical ordering step is in the time that wherein not waiting appears in arbitrary equation, after the value not waiting, new value is sorted with new value replacement.
Above-mentioned output is newly worth when obtaining new intermediate value in step and also needs this intermediate value to be assigned to central point.
Advantage of the present invention and beneficial effect are embodied in the following aspects:
1. the present invention takes into full account and utilizes the relevant information of neighbor and the feature that window moves.
2. the present invention has improved the processing speed of medium filtering.
Brief description of the drawings
Fig. 1 is execution step schematic diagram of the present invention.
Embodiment
Below in conjunction with Figure of description 1, enforcement of the present invention is further described:
A kind of image filtering processor, comprises that window moves, condition judgment, numerical ordering and new four steps of value of output.The implementation of above-mentioned steps is to the data executive condition determining step in window after carrying out window to move step, if condition is set up, export former intermediate value, otherwise carry out numerical value ordered steps, carry out afterwards the new value of output step, then enter the circulation of a new round.
The ultimate principle of medium filtering and step
Signal intermediate value is the intermediate value of arranging by signal value size order, the long one-dimensional signal for n the intermediate value of X} represents with following formula: and Med{Xt, &, Xn}, is defined as two dimensional image signal two dimension median filter device
Wherein N represents nature manifold, and A is the window that intercepts view data, and window A can have different forms, conventionally has line segment window, square window, ox-eye, cross window etc.Medium filtering is exactly the window of selecting certain forms, and it is moved on the each point of image, replaces the grey scale pixel value at window center point place with the Mesophyticum of grey scale pixel value in window.Traditional medium filtering implementation procedure is as follows:
The first step: select the window (being generally 3*3 or 5*5) of (2n+1) * (2n+1), and this window is slided along image data lines or column direction displacement;
Second step: pixel of every movement, sorts to window grey scale pixel value;
The 3rd step: the original pixels gray-scale value with sequence gained Mesophyticum for window center position;
For the image of M*N, in the situation that not considering image boundary filtering, view data needs M*N time the pixel value of window to be sorted, obviously, in the time that image data amount is larger, very time-consuming with median filtering algorithm.
The improved medium filtering implementation method of the present invention is as follows:
Because sliding window often moves past a pixel, only remove the pixel value that Far Left one is listed as, add a column data of as much at rightmost, and rest of pixels all remains unchanged simultaneously.Therefore,, in the time of the intermediate value of asking when front window, only need to consider the impact of the pixel that shifts out and the move into intermediate value on previous window, thereby avoided not changing in a large number the comparison of pixel value, thereby saved a large amount of processing times.Set about from the feature of digital picture herein, due to digital picture analog image is sampled and quantize after obtain, for image, be generally 256 grades by analog image to the quantification progression of digital picture, can see, quantized level is limited, can find out from the data of digital picture, some region at image has identical pixel value, and only have very fast variation at some transitional region pixel value, conventionally these regions are to occur in flakes in image, every width image all exists such region that is positioned at same quantized level because gray scale is close in quantizing process, consider the correlativity of image, the regionality of grey scale change, propose to improve the new algorithm of image median filter speed.
The present invention takes into full account and utilizes the relevant information of neighbor and the feature that window moves, piece image data are studied to discovery, the present invention has its foothold, the region that Digital Image Data grey scale change in original image is little, there is identical gray-scale value, in the time of the difference <T of the numerical value of two pixels, they are given to identical gray-scale value (T quantizes thresholding).
Performing step is as follows:
The first step: judge whether to meet i=l & & j=m & & k=n, if set up, export former intermediate value (having utilized the limited and neighbor of quantification number of greyscale levels to there is very large correlativity feature), otherwise enter second step;
Second step: in the time that wherein not waiting appears in arbitrary equation, replace the value not waiting by new value;
The 3rd step: new value is sorted;
The 4th step: obtain new intermediate value, this intermediate value is assigned to central point simultaneously;
The 5th step: window continues mobile, enters new round comparison.
Utilize technical solutions according to the invention, or those skilled in the art being under the inspiration of technical solution of the present invention, designs similar technical scheme, and reaching above-mentioned technique effect, is all to fall into protection scope of the present invention.
Claims (6)
1. an image filtering processor, is characterized in that: comprise that window moves, condition judgment, numerical ordering and new four steps of value of output.
2. the implementation of above-mentioned steps is, after the described window of execution moves step, the data in window are carried out to described condition judgment step, if condition is set up, export former intermediate value, otherwise carry out described numerical ordering step, carry out afterwards described output and be newly worth step, then enter the circulation of a new round.
3. a kind of image filtering processor according to claim 1, is characterized in that: it is the implementation that the window of former medium filtering moves that described window moves step.
4. a kind of image filtering processor according to claim 1, it is characterized in that: whether the condition of described condition judgment step for meeting i=l & & j=m & & k=n, if set up, utilized the limited and neighbor of quantification number of greyscale levels to there is very large correlativity feature and exported former intermediate value.
5. a kind of image filtering processor according to claim 1, is characterized in that: the implementation of described numerical ordering step, in the time that wherein not waiting appears in arbitrary equation, uses newly value after replacing the value not waiting, new value to be sorted.
6. a kind of image filtering processor according to claim 1, is characterized in that: described output is newly worth when obtaining new intermediate value in step and also needs this intermediate value to be assigned to central point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310606022.8A CN103886551A (en) | 2013-11-25 | 2013-11-25 | Image filtering processor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310606022.8A CN103886551A (en) | 2013-11-25 | 2013-11-25 | Image filtering processor |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103886551A true CN103886551A (en) | 2014-06-25 |
Family
ID=50955426
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310606022.8A Pending CN103886551A (en) | 2013-11-25 | 2013-11-25 | Image filtering processor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103886551A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105719257A (en) * | 2016-01-28 | 2016-06-29 | 河南师范大学 | Method for removing super-high-density salt-and-pepper noises of image |
| CN108304845A (en) * | 2018-01-16 | 2018-07-20 | 腾讯科技（深圳）有限公司 | Image processing method, device and storage medium |
| CN109272461A (en) * | 2018-09-04 | 2019-01-25 | 张家港江苏科技大学产业技术研究院 | Infrared image enhancing method based on median filtering and color histogram |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101908205A (en) * | 2010-06-09 | 2010-12-08 | 河北师范大学 | Magic square coding-based median filter method |
-
2013
- 2013-11-25 CN CN201310606022.8A patent/CN103886551A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101908205A (en) * | 2010-06-09 | 2010-12-08 | 河北师范大学 | Magic square coding-based median filter method |
Non-Patent Citations (1)
| Title |
|---|
| 赵高长等: "《改进的中值滤波算法在图像去噪中的应用》", 《应用光学》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105719257A (en) * | 2016-01-28 | 2016-06-29 | 河南师范大学 | Method for removing super-high-density salt-and-pepper noises of image |
| CN105719257B (en) * | 2016-01-28 | 2018-08-03 | 河南师范大学 | The drop of image ultra high density salt-pepper noise removes method |
| CN108304845A (en) * | 2018-01-16 | 2018-07-20 | 腾讯科技（深圳）有限公司 | Image processing method, device and storage medium |
| CN108304845B (en) * | 2018-01-16 | 2021-11-09 | 腾讯科技（深圳）有限公司 | Image processing method, device and storage medium |
| CN109272461A (en) * | 2018-09-04 | 2019-01-25 | 张家港江苏科技大学产业技术研究院 | Infrared image enhancing method based on median filtering and color histogram |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Single image dehazing via large sky region segmentation and multiscale opening dark channel model | |
| CN106856002B (en) | Unmanned aerial vehicle shooting image quality evaluation method | |
| CN108985169B (en) | Shop cross-door operation detection method based on deep learning target detection and dynamic background modeling | |
| CN111062293B (en) | Unmanned aerial vehicle forest flame identification method based on deep learning | |
| CN108510451B (en) | Method for reconstructing license plate based on double-layer convolutional neural network | |
| CN109064418B (en) | Non-local mean value-based non-uniform noise image denoising method | |
| CN104616290A (en) | Target detection algorithm in combination of statistical matrix model and adaptive threshold | |
| CN104048969A (en) | Tunnel defect recognition method | |
| CN102113017A (en) | System and method to generate depth data using edge detection | |
| WO2021109697A1 (en) | Character segmentation method and apparatus, and computer-readable storage medium | |
| Jin et al. | Perceptual image quality assessment using block-based multi-metric fusion (BMMF) | |
| CN105354817A (en) | Noise image automatic focusing method | |
| Priya et al. | Edge strength extraction using orthogonal vectors for shot boundary detection | |
| CN108205657A (en) | Method, storage medium and the mobile terminal of video lens segmentation | |
| CN103886551A (en) | Image filtering processor | |
| CN102663781A (en) | Sub-pixel level welding center extraction method based on visual sense | |
| Chen et al. | Impulse noise suppression with an augmentation of ordered difference noise detector and an adaptive variational method | |
| Yuhan et al. | Detection of road surface crack based on PYNQ | |
| Cui et al. | Single image haze removal based on luminance weight prior | |
| Bi et al. | Adaptive blind image restoration algorithm of degraded image | |
| Thakur et al. | Classification of color hazy images | |
| Xiao et al. | Research on 3d reconstruction technology of tool wear area | |
| CN112767446A (en) | Image tracking system for improving target tracking accuracy of infrared image tracking | |
| Hegazy et al. | Deep Image Deraining | |
| Prabavathy et al. | Gradual transition detection in shot boundary using gradual curve point |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140625 |