CN105389580B - Image rectangular neighborhood based on constant time complexity is very big or minimum value calculating method - Google Patents
Image rectangular neighborhood based on constant time complexity is very big or minimum value calculating method Download PDFInfo
- Publication number
- CN105389580B CN105389580B CN201510662958.1A CN201510662958A CN105389580B CN 105389580 B CN105389580 B CN 105389580B CN 201510662958 A CN201510662958 A CN 201510662958A CN 105389580 B CN105389580 B CN 105389580B
- Authority
- CN
- China
- Prior art keywords
- histogram
- rectangular neighborhood
- image
- multilayer
- value
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
- G06V10/507—Summing image-intensity values; Histogram projection analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/758—Involving statistics of pixels or of feature values, e.g. histogram matching
Abstract
The invention provides a kind of image rectangular neighborhood based on constant time complexity greatly (minimum) value calculating method.This method can briefly be summarised as following four step：User gives a width gray level image and is used as input picture；One three-dimensional multilayer histogram is constructed according to input picture；The histogrammic image rectangular neighborhood for calculating each layer in multilayer histogram respectively is summed, and obtains the multilayer histogram of a rectangular neighborhood summation；Rectangular neighborhood summing value according to corresponding to the multilayer histogram calculation that rectangular neighborhood is summed goes out the pixel of each in input picture is more than the level number of 0 maximum (minimum) layer, is image rectangular neighborhood greatly (minimum) value of required input picture.This method solves image rectangular neighborhood greatly (minimum) value using three-dimensional multilayer histogram and its local correlations, makes the calculating of image rectangular neighborhood greatly (minimum) value and has constant time complexity.
Description
Technical field
The present invention relates to a kind of image rectangular neighborhood greatly (minimum) value calculating method, constant time is based especially on complicated
The image rectangular neighborhood of degree greatly (minimum) value calculating method.
Background technology
The calculating of image rectangular neighborhood greatly (minimum) value, refers to seek each pixel in image by some specific algorithms
Rectangular neighborhood in very big (minimum) value.Image rectangular neighborhood greatly (minimum) value, which is calculated, to have a very wide range of applications.Example
Such as, it is accomplished by asking the maximum and minimum of image rectangular neighborhood when adjusting picture contrast.In area of pattern recognition, often need
It is required that feature extreme value in image in the rectangular neighborhood of each pixel instructs target detection or classification.Therefore image rectangle is adjacent
Domain greatly (minimum) value, which is calculated, important practical significance.
However, the method for traditional solution image rectangular neighborhood greatly (minimum) value, directly to all in rectangular neighborhood
Pixel value carries out greatly (minimum) and compared, and handles the time complexity of each pixel extreme value and does not reach constant time complexity,
But increase with the increase of rectangular neighborhood.When rectangular neighborhood than it is larger when, conventional method can consume more processing
Time, which limits its application in some algorithms.Therefore, the present invention proposes a kind of based on constant time complexity
Image rectangular neighborhood greatly (minimum) value calculating method.
The content of the invention
The purpose of the present invention：The present invention proposes a kind of image rectangular neighborhood based on constant time complexity greatly (pole
It is small) value calculating method.This method seeks image rectangular neighborhood greatly (minimum) value using three-dimensional multilayer histogram, improves calculation
The time complexity of method.
The present invention devises a kind of image rectangular neighborhood based on constant time complexity greatly (minimum) value calculating method.
This method includes following four step：
(1) gray level image of width two dimension is given as input picture I, and its width is denoted as w and h picture respectively with height
Pixel is denoted as (x, y) in element, I, then x ∈ { 0,1,2 ..., w-1 }, y ∈ { 0,1,2 ..., h-1 }.The pixel value of (x, y) is denoted as
I (x, y), for the gray level image represented by 8 bits, be up to 28=256 gray values, therefore I (x, y) ∈ 0,
1,2,...,255}.The two-dimensional rectangle that a given size is (2r+1) × (2r+1) is used as the rectangular neighborhood Ω of pixel (x, y)
(x, y) (wherein r is natural number).
(2) a three-dimensional multilayer histogram V is constructed according to input picture I in step (1).The V numbers of plies are equal to I value model
Enclose, i.e., 256 layers, every layer of histogram width and height, i.e., respectively w and h identical with I.Level number in V is represented with z, then z ∈ 0,
1,2 ..., 255 }, z layers of histogram are denoted as Vz.The all pixels of input picture are traveled through, first to value all in V according to public affairs
Formula Vz(x, y)=0 is initialized, then according to formula VI(x,y)(x, y)=1 pair V is configured.That is, for pixel
(x, y), if pixel value I (x, y) and z layers of histogram VzLevel number it is unequal, then Vz(x, y)=0, conversely, being then set to Vz
(x, y)=1.The building method only needs to constant time complexity can construction complete.
(3) the three-dimensional multilayer histogram V constructed in rectangular neighborhood Ω (x, y) in step (1) and step (2),
The histogrammic image rectangular neighborhood for calculating each layer in V respectively using the integration nomography based on constant time complexity is asked
With the multilayer histogram V' of one rectangular neighborhood summation of acquisition.
For z layers of histogram Vz, first to VzSeek the preamble of each row and be denoted as Sz.Initial value S is setz(x, -1)=
0, V is traveled through from small to largezIn per a line, according to equation below iterative calculation to VzThe preamble of each row and：
Sz(x, y)=Sz(x,y-1)+Vz(x,y)
Then, S is utilizedzIt is used as intermediate result construction histogram VzIntegrogram Az.Similarly, initial value A is setz(-1,
Y)=0, S is traveled through from small to largezIn each row, according to equation below iterative calculation to SzPer a line preamble and：
Az(x, y)=Az(x-1,y)+Sz(x,y)
Next, for z layers of histogram Vz, corresponding rectangular neighborhood summation can be efficiently calculated according to equation below
Histogram Vz'：
Vz'=Az(x-r,y-r)+Az(x+r,y+r)-Az(x-r,y+r)-Az(x+r,y-r)
Integrogram method is to seek image by a kind of the quick of Paul Viola et al. constant time complexities proposed earliest
The method of rectangular neighborhood sum.The time complexity of this method is only relevant with image size w, h and gradation of image scope, and and rectangle
Neighborhood Ω (x, y) size is unrelated, i.e., the size with r is unrelated.Referring specifically to document P.Viola and M.Jones, " Rapid
object detection using a boosted cascade of simple features,"in Proceedings
of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern
Recognition(CVPR 2001),vol.1,pp.I-511-I-518,2001。
(4) the multilayer histogram V' of the rectangular neighborhood summation in step (3), is gone out based on constant time complicated dynamic behaviour
Rectangular neighborhood summing value corresponding to the pixel of each in input picture is more than the level number of 0 maximum (minimum) layer, is required
Input picture image rectangular neighborhood greatly (minimum) value.For each pixel (x, y), rectangular neighborhood is traveled through from big to small and is asked
The multilayer histogram V' of sum each layer (even z is begun stepping through to 0 from 255), first meets V'z(x, y) ＞ 0 level number z
The image rectangular neighborhood maximum of as required input picture.Similarly, for each pixel (x, y), travel through from small to large
The multilayer histogram V' of rectangular neighborhood summation each layer (even z is begun stepping through to 255 from 0), first meets V'z(x,y)
＞ 0 level number z is the image rectangular neighborhood minimum of required input picture.The step computational methods at most only need to 256
The secondary rectangular neighborhood that can be found corresponding to some pixel greatly (minimum) value, therefore only need to constant time complexity and can count
Calculate and complete, and it is unrelated with rectangular neighborhood Ω (x, y) size, i.e., and the size with r is unrelated.
So far we have just calculated very big (minimum) value of input picture rectangular neighborhood.
Traditional method for seeking image rectangular neighborhood greatly (minimum) value, handles the time complexity of each pixel extreme value simultaneously
Do not reach constant time complexity, but increase with the increase of rectangular neighborhood size.Compared to conventional method, the present invention is related to
And the greatly beneficial effect of (minimum) value calculating method of the image rectangular neighborhood based on constant time complexity be：Three-dimensional
Multilayer histogram and its effective of spatial coherence use the operation time for causing to solve image rectangular neighborhood greatly (minimum) value
It will not increase with the increase of r values in image rectangular neighborhood Ω (x, y) size (2r+1) × (2r+1).It is given for a width
Gray level image, no matter image rectangular neighborhood is much, solves image rectangular neighborhood greatly (minimum) value of the width gray level image
Time complexity remains constant, and then effectively improves processing speed.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of embodiment.
Specific implementation method
Below by embodiment combination schematic flow sheet, the present invention is further elaborated.
Below in conjunction with the accompanying drawings to the present invention：Image rectangular neighborhood based on constant time complexity greatly (minimum) value is calculated
Method is described in detail；The present embodiment is premised on technical solution of the present invention, to be implemented, and combines detailed reality
Mode and process are applied, but protection scope of the present invention is not limited to following embodiments.
As shown in figure 1, the image rectangular neighborhood based on constant time complexity described by the present embodiment is greatly (minimum)
Value calculating method can be divided into following four step：
(1) gray level image of width two dimension is given as input picture I, and its width is denoted as w and h picture respectively with height
Pixel is denoted as (x, y) in element, I, then x ∈ { 0,1,2 ..., w-1 }, y ∈ { 0,1,2 ..., h-1 }.The pixel value of (x, y) is denoted as
I (x, y), for the gray level image represented by 8 bits, be up to 28=256 gray values, therefore I (x, y) ∈ 0,
1,2,...,255}.The two-dimensional rectangle that a given size is (2r+1) × (2r+1) is used as the rectangular neighborhood Ω of pixel (x, y)
(x, y) (wherein r is natural number).
(2) a three-dimensional multilayer histogram V is constructed according to input picture I in step (1).The V numbers of plies are equal to I value model
Enclose, i.e., 256 layers, every layer of histogram width and height, i.e., respectively w and h identical with I.Level number in V is represented with z, then z ∈ 0,
1,2 ..., 255 }, z layers of histogram are denoted as Vz.The all pixels of input picture are traveled through, first to value all in V according to public affairs
Formula Vz(x, y)=0 is initialized, then according to formula VI(x,y)(x, y)=1 pair V is configured.That is, for pixel
(x, y), if pixel value I (x, y) and z layers of histogram VzLevel number it is unequal, then Vz(x, y)=0, conversely, being then set to Vz
(x, y)=1.
(3) the three-dimensional multilayer histogram V constructed in rectangular neighborhood Ω (x, y) in step (1) and step (2),
The histogrammic image rectangular neighborhood for calculating each layer in V respectively using the integration nomography based on constant time complexity is asked
With the multilayer histogram V' of one rectangular neighborhood summation of acquisition.
For z layers of histogram Vz, first to VzSeek the preamble of each row and be denoted as Sz.Initial value S is setz(x, -1)=
0, V is traveled through from small to largezIn per a line, according to equation below iterative calculation to VzThe preamble of each row and：
Sz(x, y)=Sz(x,y-1)+Vz(x,y)
Then, S is utilizedzIt is used as intermediate result construction histogram VzIntegrogram Az.Similarly, initial value A is setz(-1,
Y)=0, S is traveled through from small to largezIn each row, according to equation below iterative calculation to SzPer a line preamble and：
Az(x, y)=Az(x-1,y)+Sz(x,y)
Next, for z layers of histogram Vz, corresponding rectangular neighborhood summation can be efficiently calculated according to equation below
Histogram Vz'：
Vz'=Az(x-r,y-r)+Az(x+r,y+r)-Az(x-r,y+r)-Az(x+r,y-r)
(4) the multilayer histogram V' of the rectangular neighborhood summation in step (3), is gone out based on constant time complicated dynamic behaviour
Rectangular neighborhood summing value corresponding to the pixel of each in input picture is more than the level number of 0 maximum (minimum) layer, is required
Input picture image rectangular neighborhood greatly (minimum) value.For each pixel (x, y), rectangular neighborhood is traveled through from big to small and is asked
The multilayer histogram V' of sum each layer (even z is begun stepping through to 0 from 255), first meets V'z(x, y) ＞ 0 level number z
The image rectangular neighborhood maximum of as required input picture.Similarly, for each pixel (x, y), travel through from small to large
The multilayer histogram V' of rectangular neighborhood summation each layer (even z is begun stepping through to 255 from 0), first meets V'z(x,y)
＞ 0 level number z is the image rectangular neighborhood minimum of required input picture.
So far we have just calculated very big (minimum) value of input picture rectangular neighborhood.
Claims (1)
1. a kind of image rectangular neighborhood based on constant time complexity is very big or minimum value calculating method, it is characterised in that：Explain
State as following four step：
(1) gives a width gray level image as input picture, gives a two-dimensional rectangle as the rectangular neighborhood of each pixel；
(2) input pictures of the in step (1) constructs in a three-dimensional multilayer histogram, multilayer histogram each layer
Histogram size is identical with input picture, and its number of plies is equal to the span of pixel value in input picture, according to input picture
Each pixel value, 0 or 1 is set to by each pixel value in multilayer histogram；
(3) the multilayer histogram in rectangular neighborhoods and step (2) of the in step (1), using based on constant time complexity
Integration nomography calculate respectively each layer in multilayer histogram histogrammic image rectangular neighborhood summation, obtain a rectangle
The multilayer histogram of neighborhood summation；
(4) the multilayer histogram of rectangular neighborhood summations of the in step (3), input figure is found based on constant time complexity
Rectangular neighborhood summing value as in corresponding to each pixel is more than 0 maximum or the level number of smallest tier, is required input
The image rectangular neighborhood of image is greatly or minimum.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510662958.1A CN105389580B (en) | 2015-10-10 | 2015-10-10 | Image rectangular neighborhood based on constant time complexity is very big or minimum value calculating method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510662958.1A CN105389580B (en) | 2015-10-10 | 2015-10-10 | Image rectangular neighborhood based on constant time complexity is very big or minimum value calculating method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105389580A CN105389580A (en) | 2016-03-09 |
| CN105389580B true CN105389580B (en) | 2017-09-12 |
Family
ID=55421850
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510662958.1A Active CN105389580B (en) | 2015-10-10 | 2015-10-10 | Image rectangular neighborhood based on constant time complexity is very big or minimum value calculating method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105389580B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105913384B (en) * | 2016-03-21 | 2017-08-15 | 温州大学 | Real-time Weighted median filtering method based on bilateral grid |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6463173B1 (en) * | 1995-10-30 | 2002-10-08 | Hewlett-Packard Company | System and method for histogram-based image contrast enhancement |
| CN102005033A (en) * | 2010-11-16 | 2011-04-06 | 中国科学院遥感应用研究所 | Method for suppressing noise by image smoothing |
| CN103606137A (en) * | 2013-11-13 | 2014-02-26 | 天津大学 | Histogram equalization method for maintaining background and detail information |
-
2015
- 2015-10-10 CN CN201510662958.1A patent/CN105389580B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6463173B1 (en) * | 1995-10-30 | 2002-10-08 | Hewlett-Packard Company | System and method for histogram-based image contrast enhancement |
| CN102005033A (en) * | 2010-11-16 | 2011-04-06 | 中国科学院遥感应用研究所 | Method for suppressing noise by image smoothing |
| CN103606137A (en) * | 2013-11-13 | 2014-02-26 | 天津大学 | Histogram equalization method for maintaining background and detail information |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105389580A (en) | 2016-03-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tian et al. | Image denoising using deep CNN with batch renormalization | |
| Tian et al. | Asymmetric CNN for image superresolution | |
| US10803554B2 (en) | Image processing method and device | |
| Wang et al. | AIPNet: Image-to-image single image dehazing with atmospheric illumination prior | |
| WO2018082185A1 (en) | Image processing method and device | |
| JP2020514890A5 (en) | ||
| Li et al. | Multi-scale single image dehazing using laplacian and gaussian pyramids | |
| CN109964250A (en) | For analyzing the method and system of the image in convolutional neural networks | |
| Sun et al. | Nir to rgb domain translation using asymmetric cycle generative adversarial networks | |
| CN105469353B (en) | The embedding grammar and device and extracting method and device of watermarking images | |
| EP2863362B1 (en) | Method and apparatus for scene segmentation from focal stack images | |
| TWI689894B (en) | Image segmentation method and apparatus | |
| CN114004754B (en) | Scene depth completion system and method based on deep learning | |
| Grogan et al. | L2 registration for colour transfer | |
| CN114627034A (en) | Image enhancement method, training method of image enhancement model and related equipment | |
| Yap et al. | A recursive soft-decision approach to blind image deconvolution | |
| CN113034358A (en) | Super-resolution image processing method and related device | |
| Zhao et al. | Restoration of motion blurred images based on rich edge region extraction using a gray-level co-occurrence matrix | |
| Wang et al. | No-reference stereoscopic image quality assessment using quaternion wavelet transform and heterogeneous ensemble learning | |
| CN109003247B (en) | Method for removing color image mixed noise | |
| CN105389580B (en) | Image rectangular neighborhood based on constant time complexity is very big or minimum value calculating method | |
| Kumar et al. | Dynamic stochastic resonance and image fusion based model for quality enhancement of dark and hazy images | |
| CN114049491A (en) | Fingerprint segmentation model training method, fingerprint segmentation device, fingerprint segmentation equipment and fingerprint segmentation medium | |
| CN108550119B (en) | Image denoising method combined with edge information | |
| Dong et al. | Shoot high-quality color images using dual-lens system with monochrome and color cameras |
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 |