CN109978876A - A kind of smog recognition methods and device based on quick bilateral filtering - Google Patents
A kind of smog recognition methods and device based on quick bilateral filtering Download PDFInfo
- Publication number
- CN109978876A CN109978876A CN201910266787.9A CN201910266787A CN109978876A CN 109978876 A CN109978876 A CN 109978876A CN 201910266787 A CN201910266787 A CN 201910266787A CN 109978876 A CN109978876 A CN 109978876A
- Authority
- CN
- China
- Prior art keywords
- weight
- points
- gaussian function
- image
- space
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20028—Bilateral filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30232—Surveillance
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of smog recognition methods based on quick bilateral filtering and device, an embodiment of the method includes: acquisition original image；Determine any vicinity points in original image in object pixel vertex neighborhood to the departure degree between the space length of target pixel points and the vicinity points pixel value and target pixel points pixel value；The space weight of the vicinity points is obtained according to the space length, the similarity weight of the vicinity points is obtained according to the departure degree, the comprehensive weight of the vicinity points is obtained according to the space weight and the similarity weight；The weighted sum that the pixel value of each vicinity points in object pixel vertex neighborhood is determined based on the comprehensive weight, the pixel value using the weighted sum as target pixel points in filtering image；Smog identification is executed in the filtering image.The embodiment is capable of providing quick bilateral filtering method and carries out image filtering to facilitate Smoke Detection.
Description
Technical field
The present invention relates to technical field area of pattern recognition more particularly to a kind of smog identifications based on quick bilateral filtering
Method and apparatus.
Background technique
Fire is common one of the major disaster of the mankind, and in order to reduce injury caused by fire, most important is exactly as early as possible
Early warning is carried out to fire.Fire fuel early period generally can not full combustion, be often accompanied by smog appearance, therefore smog can be used as
The omen that fire occurs.With industrial development, the video monitoring system based on visible light has become protection people and lives safely
Important means, and its own at low cost, high resolution, the advantage that range is wide, contactless make it be widely used in fire alarm system
System.Currently, the Smoke Detection based on video is an important research and application direction in fire alarm.
But there are many noises in original sequence, especially in smooth region, it is therefore desirable to be filtered to image
Wave not only needs smoothed image during image filtering, and needs to keep the marginal information of target in image.
Bilateral filtering is computationally intensive during image procossing, and real-time is poor, it is therefore desirable to carry out quickly bilateral
Filtering carrys out the real-time of boostfiltering.
Summary of the invention
The technical problem to be solved by the present invention is how to provide quick bilateral filtering method carries out image filtering to have
Help Smoke Detection.
In order to solve the above-mentioned technical problem, in one aspect, the present invention provides a kind of cigarettes based on quick bilateral filtering
Mist recognition methods.
The smog recognition methods based on quick bilateral filtering of the embodiment of the present invention includes: acquisition original image；It determines former
Any vicinity points in beginning image in object pixel vertex neighborhood are to the space length of target pixel points and the neighborhood pixels
Departure degree between point pixel value and target pixel points pixel value；The sky of the vicinity points is obtained according to the space length
Between weight, the similarity weight of the vicinity points is obtained according to the departure degree, according to the space weight and the phase
The comprehensive weight of the vicinity points is obtained like property weight；Each neighbour in object pixel vertex neighborhood is determined based on the comprehensive weight
The weighted sum of the pixel value of nearly pixel, the pixel value using the weighted sum as target pixel points in filtering image；And
Smog identification is executed in the filtering image.
Preferably, the space length is Euclidean distance, and the departure degree is vicinity points pixel value and target picture
The absolute value of vegetarian refreshments pixel value.
Preferably, described that the space weight of the vicinity points is obtained according to the space length, it specifically includes: will be described
Space length substitutes into Gaussian function, obtains the space weight；It is described that the vicinity points are obtained according to the departure degree
Similarity weight specifically includes: the departure degree being substituted into Gaussian function, obtains the similarity weight；According to the sky
Between weight and the similarity weight obtain the comprehensive weights of the vicinity points, specifically include: by the space weight and institute
The product of similarity weight is stated as the comprehensive weight.
Preferably, the filtering image is formed and the convolution based on two-dimensional Gaussian function；And the method is into one
Step includes: that the two-dimensional Gaussian function is converted to laterally one-dimensional Gaussian function and Vertical one dimensional Gaussian function；Utilize laterally one
Tie up Gaussian function and lateral convolution carried out to original image, recycle Vertical one dimensional Gaussian function to the image by lateral convolution into
Row longitudinal direction convolution, obtains the filtering image；Alternatively, longitudinal convolution is carried out to original image using Vertical one dimensional Gaussian function,
It recycles lateral one-dimensional Gaussian function to carry out lateral convolution to the image by longitudinal convolution, obtains the filtering image.
On the other hand, the present invention provides a kind of smog identification devices based on quick bilateral filtering.
The smog identification device based on quick bilateral filtering of the embodiment of the present invention includes: acquisition unit, for obtaining original
Beginning image；Weight calculation unit, for determining any vicinity points in original image in object pixel vertex neighborhood to target
Departure degree between the space length of pixel and the vicinity points pixel value and target pixel points pixel value；According to
The space length obtains the space weight of the vicinity points, obtains the similar of the vicinity points according to the departure degree
Property weight, the comprehensive weight of the vicinity points is obtained according to the space weight and the similarity weight；Filter unit is used
In the weighted sum for determining the pixel value of each vicinity points in object pixel vertex neighborhood based on the comprehensive weight, described will add
Power and the pixel value as target pixel points in filtering image；And recognition unit, for being executed in the filtering image
Smog identification.
Preferably, the space length is Euclidean distance, and the departure degree is vicinity points pixel value and target picture
The absolute value of vegetarian refreshments pixel value.
Preferably, weight calculation unit is further used for: the space length being substituted into Gaussian function, obtains the space
The departure degree is substituted into Gaussian function, the similarity weight is obtained, by the space weight and the similitude by weight
The product of weight is as the comprehensive weight.
Preferably, the filtering image is formed and the convolution based on two-dimensional Gaussian function；And described device is into one
Step includes: separation filter unit, for the two-dimensional Gaussian function to be converted to laterally one-dimensional Gaussian function and Vertical one dimensional height
This function；Lateral convolution is carried out to original image using laterally one-dimensional Gaussian function, recycles Vertical one dimensional Gaussian function to warp
The image for crossing lateral convolution carries out longitudinal convolution, obtains the filtering image；Alternatively, using Vertical one dimensional Gaussian function to original
Image carries out longitudinal convolution, recycles lateral one-dimensional Gaussian function to carry out lateral convolution to the image by longitudinal convolution, obtains
The filtering image.
It yet still another aspect, the present invention provides a kind of electronic equipment.
The electronic equipment of the embodiment of the present invention can include: one or more processors；Storage device, for store one or
Multiple programs；When one or more of programs are executed by one or more of processors, so that one or more of places
Manage the smog recognition methods based on quick bilateral filtering that device realizes the embodiment of the present invention.
In another aspect, the present invention provides a kind of computer readable storage medium.
The computer readable storage medium of the embodiment of the present invention, is stored thereon with computer program, and described program is processed
The smog recognition methods based on quick bilateral filtering of the embodiment of the present invention is realized when device executes.
In the technical solution of the embodiment of the present invention, target in image can be effectively kept under the premise of realizing smothing filtering
Marginal information, and two-dimensional filtering mode can be converted to linear filtering laterally and longitudinally, thus can reduce convolution fortune
Calculation amount improves image procossing real-time, realizes quick bilateral filtering.
Detailed description of the invention
Fig. 1 is the key step schematic diagram of the smog recognition methods based on quick bilateral filtering of the embodiment of the present invention；
Fig. 2 is the quick filter effect diagram of the embodiment of the present invention；
Fig. 3 is the component part schematic diagram of the smog identification device based on quick bilateral filtering of the embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiments of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Fig. 1 is the key step schematic diagram of the smog recognition methods based on quick bilateral filtering of the embodiment of the present invention.Such as
Shown in Fig. 1, the smog recognition methods based on quick bilateral filtering of the embodiment of the present invention can specifically execute following steps:
Step S101: original image is obtained.
In this step, original image can be obtained by image capture device.Original image can be gray level image, can also
To be color image.
Step S102: determine that any vicinity points in original image in object pixel vertex neighborhood arrive target pixel points
Departure degree between space length and the vicinity points pixel value and target pixel points pixel value；According to space length
The space weight of the vicinity points is obtained, the similarity weight of the vicinity points is obtained according to departure degree, according to space
Weight and similarity weight obtain the comprehensive weight of the vicinity points.
In this step, target pixel points can be the point of any pixel in original image.Above-mentioned neighborhood can be according to pre-
If rule setting, generally, neighborhood be comprising line number and the equal square area of columns.In the following description, by neighborhood
In pixel be known as being directed to the vicinity points of target pixel points, vicinity points include target pixel points.Preferably, this step
The space length of two pixels can be the Euclidean distance of plane in rapid, and the departure degree between two pixel pixel values can
To be the absolute value of pixel value.
The space length and the two pixel value that obtain any vicinity points and target pixel points departure degree it
Afterwards, space length can be substituted into Gaussian function and obtains space weight, the degree that will deviate from substitutes into Gaussian function and obtains similarity weight,
And using space weight and the product of similarity weight as comprehensive weight.It is appreciated that the space weight Jing Guo above-mentioned setting can
Reduce with the increase of space length, similarity weight can reduce with the increase of departure degree, that is, in vicinity points distance
Target pixel points farther out when, space weight is smaller；When vicinity points distance objective pixel is closer, space weight compared with
Greatly；When vicinity points and the pixel value of target pixel points are more greatly different, similarity weight is smaller；Vicinity points with
When the pixel value of target pixel points is closer to, similarity weight is larger.This property may be embodied in the comprehensive of vicinity points
It closes in weight.It should be noted that when carrying out the comparison or calculating between different pixels point, just for same grayscale mode.
For example, if original image is color image, when being based on calculated for pixel values, just for the pixel in the same channel of red, green, blue
Value.
Step S103: the weighting of the pixel value of each vicinity points in object pixel vertex neighborhood is determined based on comprehensive weight
With pixel value using weighted sum as target pixel points in filtering image.
In this step, the weighted sum that its pixel value is calculated using the comprehensive weight of each vicinity points, obtains
Weighted sum is the pixel value of target pixel points after the filtering, i.e. pixel value of the target pixel points in filtering image.
Step S104: smog identification is executed in filtering image.
In this step, preset recognizer can be executed in the image for carrying out bilateral filtering to detect smog, from
And realize the fire alarm based on smog.
Particularly, in practical application, above-mentioned filtering can be realized by the convolution based on two-dimensional Gaussian function.Due to
Two-dimensional Gaussian function is separated into the one-dimensional Gaussian function of two vertical direction, therefore two-dimensional Gaussian function is being converted to transverse direction
One-dimensional Gaussian function (horizontal axis corresponding to plane right-angle coordinate) and Vertical one dimensional Gaussian function (correspond to plane rectangular coordinates
The longitudinal axis of system) after, lateral convolution is carried out to original image using laterally one-dimensional Gaussian function, recycles Vertical one dimensional Gaussian function
The several pairs of images by lateral convolution carry out longitudinal convolution to get filtering image is arrived.Alternatively, first with Vertical one dimensional Gaussian function
Several pairs of original images carry out longitudinal convolution, and lateral one-dimensional Gaussian function is recycled to carry out lateral volume to the image by longitudinal convolution
Product, also can be obtained filtering image.
Bilateral filtering method described further below and the present invention is based on Gaussian function separation fast filtering method.
Bilateral filtering is the innovatory algorithm to gaussian filtering, have be non-iterative, local and simple characteristic, bilateral filtering
Be made of two Gaussian filters, one is applied in airspace, another is applied in gray scale codomain, can removal noise it is same
When keep marginal information.A variance is added in it in Gaussian filter function, and the variance is related with the gray value of pixel, therefore
Biggish point is differed with the pixel value during filtering to be influenced smaller, thus gaussian filtering is overcome to cause edge blurry
Disadvantage has good preservation to the edge in filter result.
Similar with gaussian filtering, bilateral filtering is also to pass through the gray value to neighborhood of pixel points based on Gaussian function
It is weighted and averaged to obtain, formula is as follows:
Wherein, I'(k) be pixel k in filtering image pixel value, k, i, N are positive integer, and I (k-i) is original graph
The pixel value of pixel (k-i) as in, W (k, i) are the comprehensive weight in bilateral filtering.
For the most common gaussian filtering, the expression formula of weight in airspace are as follows:
Wherein, Wd(k, i) is space weight, and d indicates Euclidean distance, σdFor the standard deviation of Gaussian function in airspace.
Filtering will lead to image border and occur fuzzy in airspace merely, cause filter effect bad, therefore can be by gray scale
Information introduces, and as a part in filtering, can preferably retain the marginal information in image in this way.Introduce grayscale information
Afterwards, it is as follows to can define similarity weight:
Wherein, Wr(k, i) is similarity weight, and I (k) is the pixel value of pixel k in original image, σrFor based on gray scale
The standard deviation of value.
The kernel function of bilateral filtering is the synthesis result of airspace kernel function Yu gray scale domain kernel function.In image smoothing region,
Pixel value varies less, and space right restarts main function at this time, is equivalent to carry out Gaussian Blur；In image edge area, pixel
Value variation is very big, and similarity weight plays a major role at this time, so as to keep marginal information.Comprehensive weight can be expressed as follows:
W (k, i)=Wd(k,i)*Wr(k,i)
As can be seen from the above equation, bilateral filtering can retain the marginal information in image while eliminating noise.It is bilateral
Filtering includes two mutually independent Gaussian filters, and one is applied in space, another is applied in gray scale domain.σdDetermine sky
The range that domain filter window includes, σrWeight shared by guaranteeing with target pixel points gray value to differ biggish pixel is lower,
The edge letter that differs larger compared with the weight that statuette vegetarian refreshments has with target pixel points gray value, thus can be effectively retained in image
Breath.
Bilateral filtering is computationally intensive during image procossing, and real-time is poor, needs to be promoted its real-time.Below will
It is proposed the quick bilateral filtering of increasing d type, the primarily discrete gaussian kernel function of implementation method.
Specifically, the kernel function of Gaussian function is converted into two one-dimensional functions, computational complexity can be reduced, can accelerated
Purpose, formula 1 is such as converted into following one-dimensional functions:
Wherein, HxFor laterally one-dimensional Gaussian function, HyFor Vertical one dimensional Gaussian function, σ is standard deviation, and x, y, s are positive whole
Number.
The inhibition to noise in image sequence can be realized using the method for linear filtering in this way, can reduce convolutional calculation
Amount improves real-time, realizes quick bilateral filtering.
Fig. 2 is the quick filter effect diagram of the embodiment of the present invention.Wherein, upper images are original image in Fig. 2, under
Square image is filtering image.It can be seen that filtering by quick double wave, the noise in original image is removed, and marginal information obtains
It is saved to fine.
Fig. 3 is the component part schematic diagram of the smog identification device based on quick bilateral filtering of the embodiment of the present invention.Such as
Shown in Fig. 3, the smog identification device based on quick bilateral filtering of the embodiment of the present invention can include: acquisition unit, weight calculation
Unit, filter unit and recognition unit.
Wherein, acquisition unit is for obtaining original image；Weight calculation unit is for determining object pixel in original image
Any vicinity points in vertex neighborhood are to the space length of target pixel points and the vicinity points pixel value and target picture
Departure degree between vegetarian refreshments pixel value；The space weight of the vicinity points is obtained according to the space length, according to described
Departure degree obtains the similarity weight of the vicinity points, obtains the neighbour according to the space weight and the similarity weight
The comprehensive weight of nearly pixel；Filter unit is used to determine each neighbouring picture in object pixel vertex neighborhood based on the comprehensive weight
The weighted sum of the pixel value of vegetarian refreshments, the pixel value using the weighted sum as target pixel points in filtering image；Recognition unit
For executing smog identification in the filtering image.
In embodiments of the present invention, the space length is Euclidean distance, and the departure degree is vicinity points pixel
The absolute value of value and target pixel points pixel value.
As a preferred embodiment, weight calculation unit is further used for: the space length being substituted into Gaussian function, is obtained
To the space weight, the departure degree is substituted into Gaussian function, obtains the similarity weight, by the space weight with
The product of the similarity weight is as the comprehensive weight.
Preferably, in embodiments of the present invention, the filtering image is formed and the convolution based on two-dimensional Gaussian function；
Described device further comprises separation filter unit, is used to be converted to the two-dimensional Gaussian function laterally one-dimensional Gaussian function
With Vertical one dimensional Gaussian function；Lateral convolution is carried out to original image using laterally one-dimensional Gaussian function, recycles Vertical one dimensional
Gaussian function carries out longitudinal convolution to the image by lateral convolution, obtains the filtering image；Alternatively, utilizing Vertical one dimensional height
This function carries out longitudinal convolution to original image, recycles lateral one-dimensional Gaussian function to carry out the image by longitudinal convolution horizontal
To convolution, the filtering image is obtained.
In embodiments of the present invention, a kind of electronic equipment is also provided, comprising: one or more processors and storage device.
Wherein, storage device is for storing one or more programs.When one or more of programs are by one or more of processing
Device executes, so that one or more of processors realize the smog identification side based on quick bilateral filtering of the embodiment of the present invention
Method.
On the other hand, the present invention also provides a kind of computer-readable medium, which be can be
It states included in equipment described in embodiment；It is also possible to individualism, and without in the supplying equipment.Above-mentioned calculating
Machine readable medium carries one or more program, when said one or multiple programs are executed by the equipment, so that should
The step of equipment executes includes: acquisition original image；Determine any neighborhood pixels in original image in object pixel vertex neighborhood
Point arrives the deviation journey between the space length and the vicinity points pixel value and target pixel points pixel value of target pixel points
Degree；The space weight of the vicinity points is obtained according to the space length, which is obtained according to the departure degree
The similarity weight of point, the comprehensive weight of the vicinity points is obtained according to the space weight and the similarity weight；Base
The weighted sum of the pixel value of each vicinity points in object pixel vertex neighborhood is determined in the comprehensive weight, by the weighted sum
As pixel value of the target pixel points in filtering image；Smog identification is executed in the filtering image.
In conclusion can effectively be kept under the premise of realizing smothing filtering in the technical solution of the embodiment of the present invention
The marginal information of target in image, and two-dimensional filtering mode can be converted to linear filtering laterally and longitudinally, thus may be used
Reduce convolution algorithm amount, improves image procossing real-time, realize quick bilateral filtering.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (10)
1. a kind of smog recognition methods based on quick bilateral filtering characterized by comprising
Obtain original image；
Determine any vicinity points in original image in object pixel vertex neighborhood to the space length of target pixel points and
Departure degree between the vicinity points pixel value and target pixel points pixel value；It is neighbouring that this is obtained according to the space length
The space weight of pixel obtains the similarity weight of the vicinity points according to the departure degree, according to the space right
Weight and the similarity weight obtain the comprehensive weight of the vicinity points；
The weighted sum of the pixel value of each vicinity points in object pixel vertex neighborhood is determined based on the comprehensive weight, it will be described
Pixel value of the weighted sum as target pixel points in filtering image；
And smog identification is executed in the filtering image.
2. the method according to claim 1, wherein the space length is Euclidean distance, the departure degree
For the absolute value of vicinity points pixel value and target pixel points pixel value.
3. according to the method described in claim 2, it is characterized in that,
It is described that the space weight of the vicinity points is obtained according to the space length, it specifically includes: by the space length generation
Enter Gaussian function, obtains the space weight；
It is described that the similarity weight of the vicinity points is obtained according to the departure degree, it specifically includes: by the departure degree
Gaussian function is substituted into, the similarity weight is obtained；
The comprehensive weight that the vicinity points are obtained according to the space weight and the similarity weight, specifically includes: by institute
The product of space weight and the similarity weight is stated as the comprehensive weight.
4. according to the method described in claim 3, it is characterized in that, the filtering image passes through the volume based on two-dimensional Gaussian function
It accumulates and is formed；And the method further includes:
The two-dimensional Gaussian function is converted into laterally one-dimensional Gaussian function and Vertical one dimensional Gaussian function；
Lateral convolution is carried out to original image using laterally one-dimensional Gaussian function, recycles Vertical one dimensional Gaussian function to by horizontal
Longitudinal convolution is carried out to the image of convolution, obtains the filtering image；
Alternatively, carrying out longitudinal convolution to original image using Vertical one dimensional Gaussian function, laterally one-dimensional Gaussian function pair is recycled
Image by longitudinal convolution carries out lateral convolution, obtains the filtering image.
5. a kind of smog identification device based on quick bilateral filtering characterized by comprising
Acquisition unit, for obtaining original image；
Weight calculation unit, for determining any vicinity points in original image in object pixel vertex neighborhood to object pixel
Departure degree between the space length and the vicinity points pixel value and target pixel points pixel value of point；According to described
Space length obtains the space weight of the vicinity points, is weighed according to the similitude that the departure degree obtains the vicinity points
Weight, the comprehensive weight of the vicinity points is obtained according to the space weight and the similarity weight；
Filter unit, for determining the pixel value of each vicinity points in object pixel vertex neighborhood based on the comprehensive weight
Weighted sum, the pixel value using the weighted sum as target pixel points in filtering image；And
Recognition unit, for executing smog identification in the filtering image.
6. device according to claim 5, which is characterized in that the space length is Euclidean distance, the departure degree
For the absolute value of vicinity points pixel value and target pixel points pixel value.
7. device according to claim 6, which is characterized in that
Weight calculation unit is further used for: the space length substituted into Gaussian function, obtains the space weight, it will be described
Departure degree substitutes into Gaussian function, the similarity weight is obtained, by the product of the space weight and the similarity weight
As the comprehensive weight.
8. device according to claim 7, which is characterized in that the filtering image passes through the volume based on two-dimensional Gaussian function
It accumulates and is formed；And described device further comprises:
Separation filter unit, for the two-dimensional Gaussian function to be converted to laterally one-dimensional Gaussian function and Vertical one dimensional Gaussian function
Number；Lateral convolution is carried out to original image using laterally one-dimensional Gaussian function, recycles Vertical one dimensional Gaussian function to by horizontal
Longitudinal convolution is carried out to the image of convolution, obtains the filtering image；Alternatively, using Vertical one dimensional Gaussian function to original image
Longitudinal convolution is carried out, laterally one-dimensional Gaussian function carries out transverse direction convolution to the image by longitudinal convolution to recycling, obtains described
Filtering image.
9. a kind of electronic equipment characterized by comprising
One or more processors；
Storage device, for storing one or more programs,
When one or more of programs are executed by one or more of processors, so that one or more of processors are real
The now method as described in any in claim 1-4.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that described program is processed
The method as described in any in claim 1-4 is realized when device executes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910266787.9A CN109978876A (en) | 2019-04-03 | 2019-04-03 | A kind of smog recognition methods and device based on quick bilateral filtering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910266787.9A CN109978876A (en) | 2019-04-03 | 2019-04-03 | A kind of smog recognition methods and device based on quick bilateral filtering |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109978876A true CN109978876A (en) | 2019-07-05 |
Family
ID=67082782
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910266787.9A Pending CN109978876A (en) | 2019-04-03 | 2019-04-03 | A kind of smog recognition methods and device based on quick bilateral filtering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109978876A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117423068A (en) * | 2023-12-18 | 2024-01-19 | 东莞市杰瑞智能科技有限公司 | Vehicle fire detection method and system for smart city |
| EP4276699A4 (en) * | 2021-03-04 | 2024-07-03 | Samsung Electronics Co Ltd | Image processing device and operating method therefor |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150030248A1 (en) * | 2013-07-26 | 2015-01-29 | Li-Cor, Inc. | Adaptive noise filter |
| CN104732490A (en) * | 2014-12-03 | 2015-06-24 | 南京理工大学 | CUDA-based quick bilateral filtering method |
| CN106651783A (en) * | 2016-09-29 | 2017-05-10 | 厦门美图之家科技有限公司 | Image filtering method and device |
| CN107045723A (en) * | 2017-03-10 | 2017-08-15 | 北京环境特性研究所 | Smog recognition methods based on transmissivity dynamic detection |
| CN107680054A (en) * | 2017-09-26 | 2018-02-09 | 长春理工大学 | Multisource image anastomosing method under haze environment |
| CN108648409A (en) * | 2018-04-28 | 2018-10-12 | 北京环境特性研究所 | A kind of smog detection method and device |
-
2019
- 2019-04-03 CN CN201910266787.9A patent/CN109978876A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150030248A1 (en) * | 2013-07-26 | 2015-01-29 | Li-Cor, Inc. | Adaptive noise filter |
| CN104732490A (en) * | 2014-12-03 | 2015-06-24 | 南京理工大学 | CUDA-based quick bilateral filtering method |
| CN106651783A (en) * | 2016-09-29 | 2017-05-10 | 厦门美图之家科技有限公司 | Image filtering method and device |
| CN107045723A (en) * | 2017-03-10 | 2017-08-15 | 北京环境特性研究所 | Smog recognition methods based on transmissivity dynamic detection |
| CN107680054A (en) * | 2017-09-26 | 2018-02-09 | 长春理工大学 | Multisource image anastomosing method under haze environment |
| CN108648409A (en) * | 2018-04-28 | 2018-10-12 | 北京环境特性研究所 | A kind of smog detection method and device |
Non-Patent Citations (1)
| Title |
|---|
| 王毅: "《基于偏微分方程的遥感图像处理方法》", 30 November 2016 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4276699A4 (en) * | 2021-03-04 | 2024-07-03 | Samsung Electronics Co Ltd | Image processing device and operating method therefor |
| CN117423068A (en) * | 2023-12-18 | 2024-01-19 | 东莞市杰瑞智能科技有限公司 | Vehicle fire detection method and system for smart city |
| CN117423068B (en) * | 2023-12-18 | 2024-04-02 | 东莞市杰瑞智能科技有限公司 | Vehicle fire detection method and system for smart city |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106934803B (en) | method and device for detecting surface defects of electronic device | |
| CN110246108B (en) | Image processing method, device and computer readable storage medium | |
| CN106803067B (en) | Method and device for evaluating quality of face image | |
| JP6358351B1 (en) | Surface defect inspection method and surface defect inspection apparatus | |
| CN106683100B (en) | Image segmentation defogging method and terminal | |
| US8131013B2 (en) | Method and detecting system for positions of facial parts | |
| CN111161222B (en) | Printing roller defect detection method based on visual saliency | |
| CN106934806B (en) | It is a kind of based on text structure without with reference to figure fuzzy region dividing method out of focus | |
| CN110059634B (en) | Large-scene face snapshot method | |
| CN106599783B (en) | Video occlusion detection method and device | |
| CN110717922A (en) | Image definition evaluation method and device | |
| CN109711268B (en) | Face image screening method and device | |
| US20200218343A1 (en) | Gaze point compensation method and apparatus in display device, and display device | |
| CN111145105B (en) | Image rapid defogging method and device, terminal and storage medium | |
| CN111062331B (en) | Image mosaic detection method and device, electronic equipment and storage medium | |
| CN112016469A (en) | Image processing method and device, terminal and readable storage medium | |
| CN109978876A (en) | A kind of smog recognition methods and device based on quick bilateral filtering | |
| CN115797872A (en) | Machine vision-based packaging defect identification method, system, equipment and medium | |
| CN110490886A (en) | A kind of method for automatically correcting and system for certificate image under oblique viewing angle | |
| Katramados et al. | Real-time visual saliency by division of gaussians | |
| CN109255799A (en) | A kind of method for tracking target and system based on spatially adaptive correlation filter | |
| CN111047619B (en) | Face image processing method and device and readable storage medium | |
| CN112991159B (en) | Face illumination quality evaluation method, system, server and computer readable medium | |
| CN114612418A (en) | Method, device and system for detecting surface defects of mouse shell and electronic equipment | |
| CN103226809A (en) | Image haze removal device and image haze removal method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190705 |
|
| RJ01 | Rejection of invention patent application after publication |