CN104601861B - A kind of noise-reduction method and system for optical fiber monitoring video sequence - Google Patents
A kind of noise-reduction method and system for optical fiber monitoring video sequence Download PDFInfo
- Publication number
- CN104601861B CN104601861B CN201510056904.0A CN201510056904A CN104601861B CN 104601861 B CN104601861 B CN 104601861B CN 201510056904 A CN201510056904 A CN 201510056904A CN 104601861 B CN104601861 B CN 104601861B
- Authority
- CN
- China
- Prior art keywords
- image
- mrow
- noise
- pixel
- frame
- 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
- 238000000034 method Methods 0.000 title claims abstract description 45
- 239000013307 optical fiber Substances 0.000 title claims abstract description 18
- 238000012544 monitoring process Methods 0.000 title claims abstract description 13
- 230000009467 reduction Effects 0.000 claims abstract description 53
- 238000001914 filtration Methods 0.000 claims description 32
- 230000000295 complement effect Effects 0.000 claims description 22
- 238000012545 processing Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 9
- 239000006185 dispersion Substances 0.000 claims description 8
- 238000009826 distribution Methods 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 abstract 2
- 230000033001 locomotion Effects 0.000 description 12
- 230000000694 effects Effects 0.000 description 8
- 230000005540 biological transmission Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000000877 morphologic effect Effects 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005429 filling process Methods 0.000 description 1
- 125000001475 halogen functional group Chemical group 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000021317 sensory perception Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000011282 treatment Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a kind of noise-reduction method and system for optical fiber monitoring video sequence, this noise-reduction method includes the difference image for obtaining interval two field picture, extracts the noise region included in the difference image, and generate corresponding mask image；And be filled the noise location of pixels of the mask image on first two field picture with the pixel of the correspondence position of middle two field picture, to obtain noise-reduced image.The present invention can effectively extract the nearly all graininess noise target of optical fiber monitoring video image, and movable information in original video can be kept not lose well, it will not introduce new noise with higher reliability and real-time, and present invention image after noise reduction and original picture sharpness will not be reduced.
Description
Technical Field
The invention relates to the technical field of optical fiber monitoring video signal processing, in particular to the technical field of noise reduction and enhancement of video sequences.
Background
The video monitoring has the advantages of intuition, convenience, reliability, rich information content and the like, so the video monitoring system is widely applied to occasions such as finance, commerce, traffic, residences, communities and the like, and plays a role in non-ignorable environment monitoring and safety precaution in the fields. The optical fiber transmission system has the advantages of long transmission distance, large information capacity, strong anti-interference capability, good confidentiality and the like, and becomes an important means for communication at present. The image transmission by using the optical fiber technology can effectively ensure that image data can not be stolen, and has the characteristics of high sensitivity, high resolution and the like. People focus on the aspects of high-quality pictures, high-quality vision and the like.
However, the surveillance video is inevitably polluted by noise during the processes of acquisition, transmission, storage and reproduction, and the existence of the noise not only affects the sensory perception of the video image, but also reduces the information content of the video image. In addition, video image denoising is also a preprocessing stage of video image post-processing such as compression, encoding, object recognition, and the like. Therefore, the research of the video noise reduction algorithm has very important significance. The existing video image noise reduction algorithm mainly comprises the following classification methods:
(1) the processing area of video image noise reduction can be divided into two types of pixel domain noise reduction algorithm and transform domain noise reduction algorithm: the pixel domain video image noise reduction algorithm is used for directly processing noise in a space-time three-dimensional (3D) space formed by video images, and the algorithm is early in appearance and mature in development and has the advantages of small calculation amount and good noise reduction effect; the conversion domain video image noise reduction algorithm is to convert the content of a video image, perform noise reduction processing in a conversion domain (such as a wavelet domain), and then obtain a final noise reduction video image through inverse transformation.
(2) The filtering range supported by the filter can be classified into time-domain filtering (1D filtering), spatial filtering (2D filtering), and space-time filtering (3D filtering). Compared with a two-dimensional image, the correlation of video image information exists in a one-dimensional time domain (1D) and a two-dimensional space domain (2D) at the same time, 1D filtering only utilizes the correlation of a video sequence in the time domain, 2D filtering only utilizes the correlation of a video image in the space domain, and 3D filtering simultaneously utilizes the correlation of the video image in the time domain and the space domain to perform noise reduction, so that the noise reduction effect of 3D filtering is better than that of other two types of filters. In the space-time filtering, there are two main combination modes of the time-domain filter and the spatial-domain filter: firstly, performing space-domain filtering and then performing time-domain filtering; and secondly, switching between time-domain filtering and spatial-domain filtering according to the analysis result of the image, and setting filtering strength by combining the analysis result of the image and the noise estimation size to control the time-domain filtering and the spatial-domain filtering.
(3) According to the motion estimation method, the method can be divided into two filtering methods based on motion estimation and motion self-adaption: the filtering method based on motion estimation is to fully utilize the correlation of a video image sequence in a time domain, carry out motion estimation before filtering, use a time domain filter as far as possible under the condition of not causing motion 'smear', and prove that the method can improve the noise reduction effect under most conditions; the motion adaptive filtering method is to filter in the time domain, but some adaptive mechanism is used to reduce the temporal non-stationary caused by motion. The algorithm obviously improves the visual effect after noise reduction, but the calculation amount of the noise reduction algorithm is increased due to the introduction of the motion estimation part, and the effect after the noise reduction of the video image depends on the accuracy of the motion estimation to a certain extent.
Therefore, the existing optical fiber monitoring video sequence noise reduction technology does not have a targeted high-performance algorithm at present. The traditional filtering is easy to cause the reduction of image sharpness, and the latest noise reduction technology has huge computation amount and does not have good real-time performance. Therefore, a noise reduction method with high robustness and high real-time performance is urgently needed to solve such problems.
Disclosure of Invention
The invention aims to provide a noise reduction method and a noise reduction system for an optical fiber monitoring video sequence, so that the smoothness of video transmission can be maintained on the premise of ensuring the integrity and sharpness of video information.
In order to solve the above technical problem, the present invention provides a noise reduction method for an optical fiber monitoring video sequence, which includes: obtaining a difference image of the interval frame image, extracting a noise point region contained in the difference image, and generating a corresponding mask image; and filling the noise pixel position of the mask image on the first frame image by using the pixel at the corresponding position of the intermediate frame image to obtain a noise reduction image.
Further, the method for obtaining the difference image comprises the following steps: obtaining an absolute value of a difference between adjacent frames by using a frame difference method to obtain the difference image; i.e. g (x, y) ═ gn(x,y)-gn+2(x, y) |; in the formula, gn(x, y) and gn+2(x, y) are each a continuous F-thnFrame and Fn+2A gray value at a frame image position (x, y), n being a positive integer; wherein, FnThe frame is set as the first frame, Fn+1The frame is set as an intermediate frame, Fn+2The frame is set to the last frame.
Further, the method for extracting the noise region comprises the following steps: filtering the differential image, and dynamically extracting pixels with dispersion out of range to serve as suspected noise points; and after the suspected noise point obtains a suspected noise point region through morphology, filtering the false noise point through a classifier to obtain the noise point region.
Further, the method for filtering the difference image and dynamically extracting the pixels with dispersion out of the range comprises the following steps: constructing a neighborhood window W, calculating a pixel mean value m (x, y) and a standard deviation d (x, y) in the neighborhood window, and calculating to obtain a pixel discrete tolerance threshold e (x, y) in the neighborhood window after weighting; calculating an allowable range of pixel gray scale distribution in the neighborhood window, comparing the gray scale g (x, y) of a central pixel in the neighborhood window with the allowable range, and judging that the pixel gray scale exceeds the range as a suspected noise point; i.e. the calculated allowable range of the pixel gray-scale distribution is
m(x,y)-e(x,y)≤g(x,y)≤m(x,y)+e(x,y)；
Wherein,
e(x,y)＝max(s×d(x,y),T),s≥0；
in the above formula, N represents the number of pixels in the neighborhood window, (u, v) represents the coordinates in the neighborhood window, s is the weight of the standard deviation, and T is the minimum variance.
Further, the mask image is a complementary mask image, that is, the noise pixel position of the complementary mask image is filled with the pixel at the corresponding position of the intermediate frame image on the first frame image, so as to obtain a noise reduction image.
Further, the method for generating the complementary mask image comprises the following steps:
converting the noise region into a real number type image through binarization processing; constructing an image with the same size as the original video frame, filling real number 1.0 in the pixel corresponding to the noise point position in the original video frame, and filling 0.0 in the pixel not belonging to the noise point position, that is to say
In the formula, F and B respectively represent a noise region and a non-noise region, and I is a real number type binarization noise region image; in order to prevent new noise from being introduced in the filling process and eliminate the halo effect near a noise point, smoothing the obtained real number type image consisting of 1.0 and 0.0 by using a Gaussian low-pass filter to enable the mask image 0 and the mask image 1 to be in smooth transition to obtain a mask image M; obtaining a mask image by differencing an image consisting of 1.0 and the mask image M of the same size
Further, the method for filling the noise pixel position of the complementary mask image on the first frame image by using the pixel at the corresponding position of the intermediate frame image comprises the following steps:
after obtaining the complementary mask image, FnMultiplying a frame image by a negated mask imageF thn+1And multiplying the frame image by the mask image M, and summing the two calculation result images to obtain the noise reduction image.
In another aspect, the present invention further provides a noise reduction system for an optical fiber surveillance video sequence, including:
a difference image obtaining unit for obtaining a difference image of the interval frame image; the mask generating unit is connected with the differential image acquiring unit so as to extract a noise region contained in the differential image and generate a corresponding mask image; and the pixel filling unit is connected with the mask generating unit and is suitable for filling the noise pixel position of the mask image with the pixel at the corresponding position of the intermediate frame image on the first frame image so as to obtain a noise reduction image.
Further, the noise reduction system further comprises: the mask image is a complementary mask image, the mask generation unit is also suitable for converting a noise region into a real number type image through binarization processing, and the real number type image is subjected to Gaussian low-pass filter mask image and difference calculation to obtain a negated mask image; the pixel filling unit is suitable for filling the noise pixel position of the complementary mask image with the pixel at the corresponding position of the intermediate frame image on the first frame image.
The invention has the advantages that almost all noise point targets of the video image can be effectively extracted, the motion information in the original video can be well kept not to be lost, the reliability and the real-time performance are higher, new noise can not be introduced into the image after the noise is reduced, and the sharpness of the original image can not be reduced.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 shows an algorithmic block diagram of the noise reduction method of the present invention;
FIG. 2 shows a schematic diagram of a noise extraction method;
FIG. 3 illustrates a process flow for constructing a noise-filling mask image;
FIG. 4(a) shows a schematic before processing;
FIG. 4(b) shows the schematic after noise reduction;
FIG. 5 illustrates a functional block diagram of a noise reduction system;
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic views illustrating only the basic structure of the present invention in a schematic manner, and thus show only the constitution related to the present invention.
Example 1
In the prior art, because the video is generated by continuously playing a plurality of frame images, the video is decomposed into the plurality of frame images to be correspondingly processed.
As shown in fig. 1 to 3, the noise reduction method for an optical fiber surveillance video sequence of the present invention includes: obtaining a difference image of the interval frame image, extracting a noise point region contained in the difference image, and generating a corresponding mask image; and filling (pixel filling) the noise pixel position of the mask image on the first frame image by using the pixel at the corresponding position of the intermediate frame image to obtain a noise reduction image.
The video image is restored into a denoised video image after the image frames are continuously processed by the denoising method.
Further, the method for obtaining the difference image comprises the following steps:
obtaining an absolute value of an adjacent frame difference by using a frame difference method to obtain the difference image; namely, it is
g(x,y)＝|gn(x,y)-gn+2(x,y)|；
In the formula, gn(x, y) and gn+2(x, y) are each a continuous F-thnFrame and Fn+2A gray value at a frame image position (x, y), n being a positive integer; wherein, FnThe frame is set as the first frame, Fn+2The frame is set as the last frame, Fn+1The frame is set as an intermediate frame. Wherein the difference image FdifAnd the difference of the two frames of images is obtained through absolute value calculation. In fig. 3, in order to better illustrate a specific implementation process, the first frame image (the first frame), the second frame image (the intermediate frame), and the third frame image (the last frame) are respectively denoted by F1, F2, and F3 to illustrate a specific implementation. n is a natural number, 1, 2, 3, 4 … …, and n represents the continuity of the video image.
Differential image FdifAs shown in the first row of fig. 3, there are many bright spots, i.e., noise spots that need to be removed. Meanwhile, the differential image contains moving object contours caused by movement, and the contours are important information of the video and need to be reserved.
Further, the method for extracting the noise region comprises the following steps:
firstly, a suspected noise point is extracted, namely the difference image is filtered, and pixels with dispersion out of the range are dynamically extracted to serve as the suspected noise point.
Secondly, a noise point region is obtained, namely after the suspected noise point region is obtained through morphology, the suspected noise point region is filtered through a classifier to obtain the noise point region.
Further, the specific method for extracting suspected noise points includes:
the method for filtering the differential image and dynamically extracting the pixels with dispersion out of the range comprises the following steps: constructing a neighborhood window W, calculating a pixel mean value m (x, y) and a standard deviation d (x, y) in the neighborhood window (wherein x and y represent the pixel position of an image where the center point of the neighborhood window is located, and can also be regarded as the position relation of the neighborhood window in the image), and calculating to obtain a pixel discrete tolerance threshold value e (x, y) in the neighborhood window after weighting; calculating an allowable range of pixel gray scale distribution in the neighborhood window, comparing the gray scale g (x, y) of a central pixel in the neighborhood window with the allowable range, and judging that the pixel gray scale exceeds the range as a suspected noise point; namely, it is
The calculated allowable range of the gray distribution of the pixel is
m(x,y)-e(x,y)≤g(x,y)≤m(x,y)+e(x,y)；
Wherein,
e(x,y)＝max(s×d(x,y),T),s≥0；
in the above formula, N represents the number of pixels in the neighborhood window, (u, v) represents the coordinates in the neighborhood window, s is the weight of the standard deviation, and T is the minimum variance. u is the abscissa and v is the ordinate.
Here, the neighborhood window W may be selected as 10 × 10, or 12 × 12, or 15 × 15, preferably 15 × 15, such as the neighborhood window selected as 15 × 15 in fig. 3.
The essence of image segmentation using the present invention is to segment pixels with high dispersion according to the dispersion of the gray levels in the neighborhood window, as shown in fig. 2. The neighborhood window scans the whole image according to the shape of Z, and the rapid operation can be realized only by modifying boundary data on the previous calculation result according to the mean value and variance in the neighborhood window after each displacement in the scanning process. The image processed in this step is shown in the second row of fig. 3, and all suspected noise is separated.
The specific method for obtaining the noise point region comprises the following steps:
image segmentation is performed according to the above criteria, and the segmented image is subjected to a morphological dilation operation (first step morphology) using a circular mask with a radius of 1.5, such as, but not limited to, 1.5, or 2.0, or 2.5 pixels, followed by a connected component (second morphology) by a connected operation. After the two morphological treatments, the suspected noise area can be obtained, as shown in the third row of fig. 3.
Filtering false noise points of the images subjected to the morphological processing by a classifier; specifically, a classifier, namely a support vector machine (SVM classifier), is adopted for classification and screening, and three parameters of size, brightness and circularity are used as features for classification. After being screened by the classifier, the disturbance caused by the moving object in the video can be effectively filtered, so that the effect of the video is not influenced. Since the classifier only outputs two results: noise and pseudo noise, and the classification time of a single image is in the order of milliseconds due to the extremely high real-time nature of the feature comparison simple classifiers.
As a preferred implementation manner of this embodiment, it can be seen from the third row of fig. 3 that the regions of the moving object after being classified by the support vector machine have been removed, and the rest are noisy regions. As can be seen from the local 3D image, the image region binarized by this region has only two values, 0 and 1, without excess pixels with the surrounding. Therefore, in order to enable the image to be used as a mask for pixel filling, the pixel tends to be natural, and the filled pixel is prevented from having color difference with adjacent pixels. And realizing the optimization processing of pixel filling through the complementary mask image.
The specific implementation mode comprises the following steps: in order to realize seamless transition of the boundary of the filled area, the mask image needs to be low-pass filtered to enable the edge to be in smooth transition. The mask image is a complementary mask image, namely, the noise pixel position of the complementary mask image is filled on the first frame image by using the pixel at the corresponding position of the intermediate frame image, so as to obtain a noise reduction image.
Further, the method for generating the complementary mask image comprises the following steps:
converting the noise region into a real number type image through binarization processing; namely, it is
Constructing an image with the same size as the original video frame, filling real number 1.0 in the pixel corresponding to the noise point position in the original video frame, and filling 0.0 in the pixel not belonging to the noise point position, that is to say
In the formula, F and B respectively represent a noise region and a non-noise region, and I is a real number type binarization noise region image; smoothing the obtained real number type image consisting of 1.0 and 0.0 by a Gaussian low-pass filter to obtain a mask image M;obtaining a mask image by differencing an image consisting of 1.0 and the mask image M of the same size
The method for filling the noise pixel position of the complementary mask image on the first frame image by using the pixel at the corresponding position of the intermediate frame image comprises the following steps: after obtaining the complementary mask image, FnMultiplying a frame image by a negated mask imageF thn+1And multiplying the frame image by the mask image M, and summing the two calculation result images to obtain the noise reduction image. The use of a pair of mask images converts the filling of pixels into multiplications and additions of the image, and after the filling of pixels, the noise of the intermediate frame video is effectively removed, as shown in fig. 3.
Fig. 4(a) and 4(b) show a comparison of noise reduction effects, from which the processing effect of the present invention can be clearly seen.
Example 2
In embodiment 1, the present invention further provides a noise reduction system for an optical fiber surveillance video sequence.
As shown in fig. 5, the noise reduction system for an optical fiber surveillance video sequence includes: a difference image obtaining unit for obtaining a difference image of the interval frame image; the mask generating unit is connected with the differential image acquiring unit so as to extract a noise region contained in the differential image and generate a corresponding mask image; and the pixel filling unit is connected with the mask generating unit and is suitable for filling the noise pixel position of the mask image with the pixel at the corresponding position of the intermediate frame image on the first frame image so as to obtain a noise reduction image.
Further, the mask image is a complementary mask image, the mask generation unit is further adapted to convert the noise region into a real number image through binarization processing, and the real number image is subjected to Gaussian low-pass filter mask image and difference calculation to obtain a negated mask image; the pixel filling unit is suitable for filling the noise pixel position of the complementary mask image with the pixel at the corresponding position of the intermediate frame image on the first frame image.
The detailed description of the differential image obtaining unit, the mask generating unit, and the pixel filling unit is given in embodiment 1, and is not repeated here.
The invention is especially suitable for overcoming the noise generated by the video transmitted in the form of optical fiber because the optical fiber monitoring video is an analog signal and thus the noise is generated in the transmission process.
In light of the foregoing description of the preferred embodiment of the present invention, many modifications and variations will be apparent to those skilled in the art without departing from the spirit and scope of the invention. The technical scope of the present invention is not limited to the content of the specification, and must be determined according to the scope of the claims.
Claims (6)
1. A noise reduction method for an optical fiber monitoring video sequence comprises the following steps:
obtaining a difference image of the interval frame image, extracting a noise point region contained in the difference image, and generating a corresponding mask image; and
filling the noise pixel position of the mask image on the first frame image by using the pixel at the corresponding position of the intermediate frame image to obtain a noise reduction image;
the method for obtaining the difference image comprises the following steps:
obtaining an absolute value of an adjacent frame difference by using a frame difference method to obtain the difference image; namely, it is
g(x,y)＝|gn(x,y)-gn+2(x,y)|；
In the formula, gn(x, y) and gn+2(x, y) are each a continuous F-thnFrame and Fn+2A gray value at a frame image position (x, y), n being a positive integer; wherein, FnThe frame is set as the first frame, Fn+1The frame is set as an intermediate frame, Fn+2Setting the frame as a last frame;
the method for extracting the noise region comprises the following steps:
filtering the differential image, and dynamically extracting pixels with dispersion out of range to serve as suspected noise points; and
after the suspected noise point is subjected to morphology to obtain a suspected noise point region, filtering out false noise points through a classifier to obtain the noise point region;
the method for filtering the differential image and dynamically extracting the pixels with dispersion out of the range comprises the following steps:
constructing a neighborhood window W, calculating a pixel mean value m (x, y) and a standard deviation d (x, y) in the neighborhood window, and calculating to obtain a pixel discrete tolerance threshold e (x, y) in the neighborhood window after weighting; and
calculating an allowable range of pixel gray scale distribution in a neighborhood window, comparing the gray scale g (x, y) of a central pixel in the neighborhood window with the allowable range, and judging that the pixel gray scale exceeds the range as a suspected noise point; namely, it is
The calculated allowable range of the gray distribution of the pixel is
m(x,y)-e(x,y)≤g(x,y)≤m(x,y)+e(x,y)；
Wherein,
<mrow> <mi>m</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mo>&Sigma;</mo> <mrow> <mi>u</mi> <mo>,</mo> <mi>v</mi> </mrow> </munder> <mi>g</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>,</mo> <mi>v</mi> <mo>)</mo> </mrow> </mrow> <mi>N</mi> </mfrac> <mo>;</mo> </mrow>
<mrow> <mi>d</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>=</mo> <msqrt> <mfrac> <mrow> <munder> <mo>&Sigma;</mo> <mrow> <mi>u</mi> <mo>,</mo> <mi>v</mi> </mrow> </munder> <msup> <mrow> <mo>&lsqb;</mo> <mi>g</mi> <mrow> <mo>(</mo> <mi>u</mi> <mo>,</mo> <mi>v</mi> <mo>)</mo> </mrow> <mo>-</mo> <mi>m</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>&rsqb;</mo> </mrow> <mn>2</mn> </msup> </mrow> <mi>N</mi> </mfrac> </msqrt> <mo>;</mo> </mrow>
e(x,y)＝max(s×d(x,y),T),s≥0；
in the above formula, N represents the number of pixels in the neighborhood window, (u, v) represents the coordinates in the neighborhood window, s is the weight of the standard deviation, and T is the minimum variance.
2. The noise reduction method according to claim 1,
the mask image being a complementary mask image, i.e.
And filling the noise pixel position of the complementary mask image on the first frame image by using the pixel at the corresponding position of the intermediate frame image to obtain a noise reduction image.
3. The noise reduction method according to claim 2, wherein the generation method of the complementary mask image includes:
converting the noise region into a real number type image through binarization processing; namely, it is
Constructing an image with the same size as the original video frame, filling real number 1.0 in the pixel corresponding to the noise point position in the original video frame, and filling 0.0 in the pixel not belonging to the noise point position, that is to say
<mrow> <mi>I</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>1.0</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <mi>p</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>&Element;</mo> <mi>F</mi> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mn>0.0</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <mi>p</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>&Element;</mo> <mi>B</mi> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>;</mo> </mrow>
In the formula, F and B respectively represent a noise region and a non-noise region, and I is a real number type binarization noise region image;
smoothing the obtained real number type image consisting of 1.0 and 0.0 by a Gaussian low-pass filter to obtain a mask image M; obtaining a mask image by differencing an image consisting of 1.0 and the mask image M of the same size
4. The method of reducing noise according to claim 3, wherein the filling of the noisy pixel position of the complementary mask image with the pixel at the corresponding position of the intermediate frame image on the first frame image comprises:
after obtaining the complementary mask image, FnMultiplying a frame image by a negated mask imageF thn+1And multiplying the frame image by the mask image M, and summing the two calculation result images to obtain the noise reduction image.
5. A noise reduction system applying the noise reduction method for the optical fiber surveillance video sequence according to claim 1, comprising:
a difference image obtaining unit for obtaining a difference image of the interval frame image;
the mask generating unit is connected with the differential image acquiring unit so as to extract a noise region contained in the differential image and generate a corresponding mask image; and
and the pixel filling unit is connected with the mask generating unit and is suitable for filling the noise pixel position of the mask image with the pixel at the corresponding position of the intermediate frame image on the first frame image so as to obtain a noise reduction image.
6. The noise reduction system of claim 5,
the mask image is a complementary mask image,
the mask generation unit is also suitable for converting the noise region into a real number type image through binarization processing, and processing the real number type image through a Gaussian low-pass filter mask image and calculating the difference to obtain a negated mask image;
the pixel filling unit is suitable for filling the noise pixel position of the complementary mask image with the pixel at the corresponding position of the intermediate frame image on the first frame image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510056904.0A CN104601861B (en) | 2015-02-03 | 2015-02-03 | A kind of noise-reduction method and system for optical fiber monitoring video sequence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510056904.0A CN104601861B (en) | 2015-02-03 | 2015-02-03 | A kind of noise-reduction method and system for optical fiber monitoring video sequence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104601861A CN104601861A (en) | 2015-05-06 |
| CN104601861B true CN104601861B (en) | 2017-08-29 |
Family
ID=53127332
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510056904.0A Expired - Fee Related CN104601861B (en) | 2015-02-03 | 2015-02-03 | A kind of noise-reduction method and system for optical fiber monitoring video sequence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104601861B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109242782B (en) * | 2017-07-11 | 2022-09-09 | 深圳市道通智能航空技术股份有限公司 | Noise processing method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102014240A (en) * | 2010-12-01 | 2011-04-13 | 深圳市蓝韵实业有限公司 | Real-time medical video image denoising method |
| CN102263885A (en) * | 2010-05-27 | 2011-11-30 | 于培宁 | Method for denoising image sequence |
| CN102663696A (en) * | 2012-03-31 | 2012-09-12 | 广东威创视讯科技股份有限公司 | Denoising method of enlarged image and system thereof |
| CN103179325A (en) * | 2013-03-26 | 2013-06-26 | 北京理工大学 | Self-adaptive 3D (Three-Dimensional) noise reduction method for low signal-to-noise ratio video under fixed scene |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7199838B2 (en) * | 2004-06-17 | 2007-04-03 | Samsung Electronics Co., Ltd. | Motion adaptive noise reduction apparatus and method for video signals |
-
2015
- 2015-02-03 CN CN201510056904.0A patent/CN104601861B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102263885A (en) * | 2010-05-27 | 2011-11-30 | 于培宁 | Method for denoising image sequence |
| CN102014240A (en) * | 2010-12-01 | 2011-04-13 | 深圳市蓝韵实业有限公司 | Real-time medical video image denoising method |
| CN102663696A (en) * | 2012-03-31 | 2012-09-12 | 广东威创视讯科技股份有限公司 | Denoising method of enlarged image and system thereof |
| CN103179325A (en) * | 2013-03-26 | 2013-06-26 | 北京理工大学 | Self-adaptive 3D (Three-Dimensional) noise reduction method for low signal-to-noise ratio video under fixed scene |
Non-Patent Citations (1)
| Title |
|---|
| 基于三帧差法的运动目标检测方法研究;赵健;《中国优秀硕士学位论文全文数据库 信息科技辑 》;20140115;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104601861A (en) | 2015-05-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Inter/intra-category discriminative features for aerial image classification: A quality-aware selection model | |
| CN109636766B (en) | Edge information enhancement-based polarization difference and light intensity image multi-scale fusion method | |
| WO2021129569A1 (en) | Human action recognition method | |
| CN108564597B (en) | Video foreground object extraction method fusing Gaussian mixture model and H-S optical flow method | |
| Jia et al. | A two-step approach to see-through bad weather for surveillance video quality enhancement | |
| CN111709888B (en) | Aerial image defogging method based on improved generation countermeasure network | |
| CN103020898B (en) | Sequence iris image super resolution ratio reconstruction method | |
| CN103295225B (en) | Train bogie edge detection method under the conditions of low-light | |
| CN110349191A (en) | A kind of visual tracking method based on perceptual hash algorithm | |
| Hsieh et al. | Fast and robust infrared image small target detection based on the convolution of layered gradient kernel | |
| Das et al. | A comparative study of single image fog removal methods | |
| Pan et al. | Single-image dehazing via dark channel prior and adaptive threshold | |
| Singh et al. | Gaussian and Laplacian of Gaussian weighting functions for robust feature based tracking | |
| CN102222321A (en) | Blind reconstruction method for video sequence | |
| CN110047041A (en) | A kind of empty-frequency-domain combined Traffic Surveillance Video rain removing method | |
| CN107748885B (en) | Method for recognizing fuzzy character | |
| CN113902694A (en) | Target detection method based on dynamic and static combination | |
| CN113129214A (en) | Super-resolution reconstruction method based on generation countermeasure network | |
| CN108596928A (en) | Based on the noise image edge detection method for improving Gauss-Laplace operator | |
| CN104601861B (en) | A kind of noise-reduction method and system for optical fiber monitoring video sequence | |
| CN111147804B (en) | Video frame reconstruction method | |
| Hu et al. | Maritime video defogging based on spatial-temporal information fusion and an improved dark channel prior | |
| CN107798676B (en) | No-reference image quality perception method established based on multi-scale natural scene statistics | |
| CN110633705A (en) | Low-illumination imaging license plate recognition method and device | |
| Chen et al. | A study of image segmentation algorithms combined with different image preprocessing methods for thyroid ultrasound images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| 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: 20170829 |