CN107767387A - Profile testing method based on the global modulation of changeable reception field yardstick - Google Patents
Profile testing method based on the global modulation of changeable reception field yardstick Download PDFInfo
- Publication number
- CN107767387A CN107767387A CN201711098829.XA CN201711098829A CN107767387A CN 107767387 A CN107767387 A CN 107767387A CN 201711098829 A CN201711098829 A CN 201711098829A CN 107767387 A CN107767387 A CN 107767387A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- pixel
- msup
- 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.)
- Granted
Links
- 238000012360 testing method Methods 0.000 title claims abstract description 15
- 238000010606 normalization Methods 0.000 claims abstract description 24
- 230000005764 inhibitory process Effects 0.000 claims abstract description 23
- 230000004044 response Effects 0.000 claims abstract description 23
- 238000001914 filtration Methods 0.000 claims abstract description 22
- 230000001629 suppression Effects 0.000 claims description 7
- 230000009977 dual effect Effects 0.000 claims description 4
- 238000000034 method Methods 0.000 abstract description 13
- 230000000694 effects Effects 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 31
- 238000001514 detection method Methods 0.000 description 6
- 230000009286 beneficial effect Effects 0.000 description 2
- 208000003098 Ganglion Cysts Diseases 0.000 description 1
- 208000005400 Synovial Cyst Diseases 0.000 description 1
- 238000005267 amalgamation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- 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
Abstract
The present invention is intended to provide a kind of profile testing method based on the global modulation of changeable reception field yardstick, comprises the following steps:A, image to be detected through gray proces is inputted, the normalization difference of Gaussian filter value of each pixel is calculated;B, high yardstick value, low scale-value, the threshold value of scaling function are preset, the normalization difference of Gaussian filter value of each pixel is contrasted with threshold value respectively, determines the scaling function value of each pixel;C, inhibition strength and respectively the multiple directions parameter of circumference are preset;Gabor filtering is carried out according to all directions parameter respectively to each pixel in image to be detected, calculates the classical receptive field stimuli responsive of each pixel;D, the inhibition response of each pixel is calculated;The profile that the classical receptive field stimuli responsive of each pixel and inhibition response are calculated to the pixel responds and handles to obtain the final profile value of each pixel E, and then obtains final profile figure.This method has the characteristics of simulated effect is good, outline identification rate is high.
Description
Technical field
The present invention relates to Computer Image Processing field, and in particular to a kind of based on the global modulation of changeable reception field yardstick
Profile testing method.
Background technology
Contour detecting is a basic task of computer vision field, changes institute different from being defined as strong brightness
The edge of sign, profile generally represent a target to the border of other targets.Improve the basic skills of contour detecting performance just
It is the information of amalgamation of global, in order to improve the performance of contour detecting model, what many researchers did the best goes to original detection calculation
Son and suppression model are improved;Based on Scale-space theory, the size of the corresponding one group of neuron receptive field of each yardstick, god
There is the characteristic under different scale through the different receptive field size of ganglion cell;Therefore, receptive field model yardstick is considered in a model
Can be as the developing direction in the field.
The content of the invention
The present invention is intended to provide a kind of profile testing method based on the global modulation of changeable reception field yardstick, this method have
The characteristics of simulated effect is good, outline identification rate is high.
Technical scheme is as follows:
A, image to be detected through gray proces is inputted, height is carried out using difference of Gaussian function to the gray value of each pixel
This differential filtering, obtain the difference of Gaussian filter value of each pixel, respectively in the difference of Gaussian filter value of each pixel just
Value and negative value are normalized, and obtain the normalization difference of Gaussian filter function of each pixel;Utilize returning for each pixel
One, which changes gray value of the difference of Gaussian filter function respectively with corresponding pixel points, carries out convolution, obtains the normalization Gauss of each pixel
Differential filtering value;
B, high yardstick value, the low scale-value of scaling function, predetermined threshold value, by the normalization difference of Gaussian of each pixel are preset
Filter value is contrasted with threshold value respectively, if pixel normalization difference of Gaussian filter value is more than or equal to threshold value, the picture
Scaling function value corresponding to vegetarian refreshments is high yardstick value, if pixel normalization difference of Gaussian filter value is less than threshold value, the picture
Scaling function value corresponding to vegetarian refreshments is low scale-value;
C, inhibition strength is preset, presets the multiple directions parameter for dividing equally circumference;To each pixel in image to be detected point
Gabor filtering is not carried out according to all directions parameter, obtains the response of all directions of each pixel, institute in wherein Gabor filtering
The scaling function value used is each pixel identified high yardstick value or low scale-value in stepb;For each pixel,
The maximum in the response of its all directions is chosen, the classical receptive field stimuli responsive as the pixel;
D, by the distance weighting function of the classical receptive field stimuli responsive of each pixel and the pixel carry out after convolution with
Scale Multiplication corresponding to the pixel, obtain the inhibition response of each pixel;
E, the classical receptive field stimuli responsive of each pixel is subtracted into the inhibition response of the pixel and multiplying for inhibition strength
Product, the profile response of the pixel is obtained, profile response is handled using non-maxima suppression and dual threshold, obtains each pixel
Final profile value, and then obtain final profile figure.
Preferably, described step A is specially:
Described difference of Gaussian function DoGσ(x, y) is:
Wherein σ is initialization yardstick;
Described normalization difference of Gaussian filter function wσ(x, y) is:
Described normalization difference of Gaussian filter value woutσ(x, y) is:
woutσ(x, y)=I (x, y) * wσ(x,y) (3)。
Preferably, described step B is specially:
Described scaling function σ (x, y) is:
Wherein σHFor high yardstick value, σLFor low scale-value, σH=σL+ s, t be level off to 0 threshold value, s be yardstick step-length.
Preferably, described step C is specially:
The two-dimensional Gabor function expression of described Gabor filter group is as follows:
Whereinγ is one and represents oval receptive field major and minor axis ratio
Constant, parameter lambda is wavelength, and σ (x, y) is scaling function, and 1/ λ is the spatial frequency of cosine function,It is phase angular dimensions, θ is
The directioin parameter of Gabor filtering;
I (x, y) is the gray value of each pixel of image to be detected, and * is convolution operator;
Gabor energy values are calculated as follows:
Wherein θiFor a direction of Gabor filtering, NθFor the number in the direction of Gabor filtering;
Described classical receptive field stimuli responsive Ec (x, y;σ (x, y)) be:
Preferably, described step D is specially:
Described distance weighting function wσ(x, y) is:
Wherein,
Wherein, | | | |1For (L1) norm, H (x)=max (0, x), DoG (x, y) they are expression formula corresponding to DoG templates;
The inhibition response Inh (x, y) of each pixel is:
Inh (x, y)=Ec (x, y;σ(x,y))*wσ(x,y)σ(x,y) (11)。
Preferably, described step E is specially:
Described profile responds R (x, y):
R (x, y)=H (Ec (x, y;σ(x,y))-αInh(x,y)) (12);
Wherein H (x)=max (0, x), α are inhibition strength.
Differentiation selection of the present invention to high and low scale-value is beneficial to extraction profile based on high yardstick value, low scale-value is beneficial to suppression
The principle of texture processed, therefore judged according to the normalized difference of Gaussian filter value of each pixel, if the filter value is noticeably greater than
0, then the pixel is likely on the position of profile, therefore profile is extracted from high yardstick value;If otherwise filtering
For value near 0, then the pixel is likely on the position of texture, therefore texture is suppressed from low scale-value;Will
Difference of Gaussian is normalized so that the weight of positive and negative filter value is consistent when being filtered, and reduces error recognition rate;Cause
This, is profile or texture with reference to corresponding to above-mentioned filter value judges the pixel, so as to select corresponding high yardstick value or low
Scale-value carries out the calculating of inhibition response, and the suppression of texture can be also taken into account while ensureing and not missing profile, has been avoided
There is excessive useless texture, influence the effect of outline identification;Also, suppress to ring to calculate by combining distance weighting function
Should, it can preferably reflect the rejection characteristic of non-classical receptive field, improve contour detecting rate.
In summary, profile testing method of the invention both maintained the integrality of profile and also greatly eliminate it is unnecessary
Grain background, and more meet the wild spatial frequency characteristic of visual experience.
Brief description of the drawings
Fig. 1 is the contour detecting comparison diagram of the method for the embodiment of the present invention 1 and the method for document 1.
Embodiment
The present invention is illustrated with reference to the accompanying drawings and examples.
Embodiment 1
The profile testing method based on the global modulation of changeable reception field yardstick that the present embodiment provides, comprises the following steps:
A, image to be detected through gray proces is inputted, height is carried out using difference of Gaussian function to the gray value of each pixel
This differential filtering, obtain the difference of Gaussian filter value of each pixel, respectively in the difference of Gaussian filter value of each pixel just
Value and negative value are normalized, and obtain the normalization difference of Gaussian filter function of each pixel;Utilize returning for each pixel
One, which changes gray value of the difference of Gaussian filter function respectively with corresponding pixel points, carries out convolution, obtains the normalization Gauss of each pixel
Differential filtering value;
Described difference of Gaussian function DoGσ(x, y) is:
Wherein σ is initialization yardstick;
Described normalization difference of Gaussian filter function wσ(x, y) is:
Described normalization difference of Gaussian filter value woutσ(x, y) is:
woutσ(x, y)=I (x, y) * wσ(x,y) (3);
B, high yardstick value, the low scale-value of scaling function, predetermined threshold value, by the normalization difference of Gaussian of each pixel are preset
Filter value is contrasted with threshold value respectively, if pixel normalization difference of Gaussian filter value is more than or equal to threshold value, the picture
Scaling function value corresponding to vegetarian refreshments is high yardstick value, if pixel normalization difference of Gaussian filter value is less than threshold value, the picture
Scaling function value corresponding to vegetarian refreshments is low scale-value;
Described step B is specially:
Described scaling function σ (x, y) is:
Wherein σHFor high yardstick value, σLFor low scale-value, σH=σL+ s, t be level off to 0 threshold value, s be yardstick step-length;
C, inhibition strength is preset, presets the multiple directions parameter for dividing equally circumference;To each pixel in image to be detected point
Gabor filtering is not carried out according to all directions parameter, obtains the response of all directions of each pixel, institute in wherein Gabor filtering
The scaling function value used is each pixel identified high yardstick value or low scale-value in stepb;For each pixel,
The maximum in the response of its all directions is chosen, the classical receptive field stimuli responsive as the pixel;
The two-dimensional Gabor function expression of described Gabor filter group is as follows:
Whereinγ is one and represents oval receptive field major and minor axis ratio
Constant, parameter lambda is wavelength, and σ (x, y) is scaling function, and 1/ λ is the spatial frequency of cosine function,It is phase angular dimensions, θ is
The directioin parameter of Gabor filtering;
I (x, y) is the gray value of each pixel of image to be detected, and * is convolution operator;
Gabor energy values are calculated as follows:
Wherein θiFor a direction of Gabor filtering, NθFor the number in the direction of Gabor filtering;
Described classical receptive field stimuli responsive Ec (x, y;σ (x, y)) be:
D, by the distance weighting function of the classical receptive field stimuli responsive of each pixel and the pixel carry out after convolution with
Scale Multiplication corresponding to the pixel, obtain the inhibition response of each pixel;
Described step D is specially:
Described distance weighting function wσ(x, y) is:
Wherein,
Wherein, | | | |1For (L1) norm, H (x)=max (0, x), DoG (x, y) they are expression formula corresponding to DoG templates;
The inhibition response Inh (x, y) of each pixel is:
Inh (x, y)=Ec (x, y;σ(x,y))*wσ(x,y)σ(x,y) (11);
E, the classical receptive field stimuli responsive of each pixel is subtracted into the inhibition response of the pixel and multiplying for inhibition strength
Product, the profile response of the pixel is obtained, profile response is handled using non-maxima suppression and dual threshold, obtains each pixel
Final profile value, and then obtain final profile figure;
Described step E is specially:
Described profile responds R (x, y):
R (x, y)=H (Ec (x, y;σ(x,y))-αInh(x,y)) (12);
Wherein H (x)=max (0, x), α are inhibition strength.
The contour detecting isotropic model for below providing the profile testing method of the present embodiment and document 1 carries out effective
Property contrast, document 1 is as follows:
Document 1:Cosmin Grigorescu,Nicolai Petkov,and Michel
A.Westenberg.Contour Detection Based on Nonclassical Receptive Field
Inhibition[J].IEEE Transactions on image processing,vol.12,no.7,july 2003
729-739;
To ensure the validity of contrast, used and identical non-maxima suppression and dual threashold in document 1 for the present embodiment
Value processing, wherein the two threshold value t includedh,tlIt is arranged to tl=0.5th, calculated by threshold value quantile p and obtained;
This Performance Evaluating Indexes F uses the following standard provided in document 2:
Document 2 is " D.R.Martin, C.C.Fowlkes, and J.Malik, " Learning to detect
natural image boundaries using local brightness,color,and texture cues,"IEEE
transactions on pattern analysis and machine intelligence,vol.26,pp.530-549,
2004.”;
P represents accuracy in formula, and R represents coverage rate, and evaluation metricses P values represent profile between [0,1], closer to 1
The effect of detection is better;
Choose 3 secondary classic map pictures shown in Fig. 1 and carry out Usefulness Pair ratio, method and reality in document 1 is respectively adopted
Apply the method for example 1 and contour detecting is carried out to above-mentioned 3 width figure, the parameter group of the wherein method selection of embodiment 1 is as shown in table 1;
The parameter group table of 1 embodiment of table 1
Method in document 1 uses following 80 groups of parameters:α={ 1.0,1.2 },
σ={ 1.4,1.6,1.8,2.0,2.2,2.4,2.6,2.8 }, p={ 0.5,0.4,0.3,0.2,0.1 };
It is as shown in Figure 1 the artwork of 3 secondary classic map pictures, TP figure, the optimal profile of the method for document 1 detection, implements
The optimal profile of the method for example 1 detection;It is as shown in table 2 the optimal F values that the method for document 1 of above-mentioned 3 width image detects and implementation
The optimal F values contrast of the method for example 1 detection;
The F value contrast tables of table 2
No matter it can be seen from the results above that from the effect of contours extract or from performance indications parameter, implement
The method of example 1 is superior to the profile testing method in document 1.
Claims (6)
1. a kind of profile testing method based on the global modulation of changeable reception field yardstick, it is characterised in that comprise the following steps:
A, image to be detected through gray proces is inputted, Gaussian difference is carried out using difference of Gaussian function to the gray value of each pixel
Point filtering, obtain the difference of Gaussian filter value of each pixel, respectively in the difference of Gaussian filter value of each pixel on the occasion of with
Negative value is normalized, and obtains the normalization difference of Gaussian filter function of each pixel;Utilize the normalization of each pixel
Difference of Gaussian filter function carries out convolution with the gray value of corresponding pixel points respectively, obtains the normalization difference of Gaussian of each pixel
Filter value;
B, high yardstick value, the low scale-value of scaling function are preset, predetermined threshold value, the normalization difference of Gaussian of each pixel is filtered
Value is contrasted with threshold value respectively, if pixel normalization difference of Gaussian filter value is more than or equal to threshold value, the pixel
Corresponding scaling function value is high yardstick value, if pixel normalization difference of Gaussian filter value is less than threshold value, the pixel
Corresponding scaling function value is low scale-value;
C, inhibition strength is preset, presets the multiple directions parameter for dividing equally circumference;Each pixel in image to be detected is pressed respectively
Gabor filtering is carried out according to all directions parameter, obtains the responses of all directions of each pixel, employed in wherein Gabor filtering
Scaling function value be each pixel identified high yardstick value or low scale-value in stepb;For each pixel, choose
Maximum in the response of its all directions, the classical receptive field stimuli responsive as the pixel;
D, by the distance weighting function of the classical receptive field stimuli responsive of each pixel and the pixel carry out after convolution with the picture
Scale Multiplication corresponding to vegetarian refreshments, obtain the inhibition response of each pixel;
E, the classical receptive field stimuli responsive of each pixel is subtracted into the inhibition response of the pixel and the product of inhibition strength, obtained
Profile to the pixel is responded, and profile response is handled using non-maxima suppression and dual threshold, obtains each pixel most
Whole profile value, and then obtain final profile figure.
2. the profile testing method as claimed in claim 1 based on the global modulation of changeable reception field yardstick, it is characterised in that:
Described step A is specially:
Described difference of Gaussian function DoGσ(x, y) is:
<mrow>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>2</mn>
<mi>&pi;</mi>
<msup>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mi>&sigma;</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mfrac>
<mrow>
<msup>
<mi>x</mi>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mi>y</mi>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mi>&sigma;</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>2</mn>
<msup>
<mi>&pi;&sigma;</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mfrac>
<mrow>
<msup>
<mi>x</mi>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mi>y</mi>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<mi>&sigma;</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein σ is initialization yardstick;
Described normalization difference of Gaussian filter function wσ(x, y) is:
<mrow>
<msub>
<mi>w</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mfrac>
<mrow>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo><</mo>
<mn>0</mn>
</mrow>
</msub>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>,</mo>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo><</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mfrac>
<mrow>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>&GreaterEqual;</mo>
<mn>0</mn>
</mrow>
</msub>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>,</mo>
<msub>
<mi>DoG</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>&GreaterEqual;</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Described normalization difference of Gaussian filter value woutσ(x, y) is:
woutσ(x, y)=I (x, y) * wσ(x,y) (3)。
3. the profile testing method as claimed in claim 2 based on the global modulation of changeable reception field yardstick, it is characterised in that:
Described step B is specially:
Described scaling function σ (x, y) is:
<mrow>
<mi>&sigma;</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>&sigma;</mi>
<mi>L</mi>
</msub>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>wout</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo><</mo>
<mi>t</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>&sigma;</mi>
<mi>H</mi>
</msub>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>wout</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>&GreaterEqual;</mo>
<mi>t</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein σHFor high yardstick value, σLFor low scale-value, σH=σL+ s, t be level off to 0 threshold value, s be yardstick step-length.
4. the profile testing method as claimed in claim 3 based on the global modulation of changeable reception field yardstick, it is characterised in that:
Described step C is specially:
The two-dimensional Gabor function expression of described Gabor filter group is as follows:
Whereinγ is one and represents the normal of oval receptive field major and minor axis ratio
Number, parameter lambda are wavelength, and σ (x, y) is scaling function, and 1/ λ is the spatial frequency of cosine function,It is phase angular dimensions, θ Gabor
The directioin parameter of filtering;
I (x, y) is the gray value of each pixel of image to be detected, and * is convolution operator;
Gabor energy values are calculated as follows:
<mrow>
<msub>
<mi>E</mi>
<mrow>
<mi>&lambda;</mi>
<mo>,</mo>
<mi>&sigma;</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>i</mi>
</msub>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msqrt>
<mrow>
<msub>
<msup>
<mi>e</mi>
<mn>2</mn>
</msup>
<mrow>
<mi>&lambda;</mi>
<mo>,</mo>
<mi>&sigma;</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<mn>0</mn>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<msup>
<mi>e</mi>
<mn>2</mn>
</msup>
<mrow>
<mi>&lambda;</mi>
<mo>,</mo>
<mi>&sigma;</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<mi>&pi;</mi>
<mo>/</mo>
<mn>2</mn>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</msqrt>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
<mrow>
<msub>
<mi>&theta;</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<mi>&pi;</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
<msub>
<mi>N</mi>
<mi>&theta;</mi>
</msub>
</mfrac>
<mo>,</mo>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mo>...</mo>
<msub>
<mi>N</mi>
<mi>&theta;</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein θiFor a direction of Gabor filtering, NθFor the number in the direction of Gabor filtering;
Described classical receptive field stimuli responsive Ec (x, y;σ (x, y)) be:
<mrow>
<mi>E</mi>
<mi>c</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>;</mo>
<mi>&sigma;</mi>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
<mo>{</mo>
<msub>
<mi>E</mi>
<mrow>
<mi>&lambda;</mi>
<mo>,</mo>
<mi>&sigma;</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>i</mi>
</msub>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>|</mo>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mo>...</mo>
<msub>
<mi>N</mi>
<mi>&theta;</mi>
</msub>
<mo>}</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>9</mn>
<mo>)</mo>
</mrow>
<mo>.</mo>
</mrow>
5. the profile testing method as claimed in claim 4 based on the global modulation of changeable reception field yardstick, it is characterised in that:
Described step D is specially:
Described distance weighting function wσ(x, y) is:
<mrow>
<msub>
<mi>w</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mi>H</mi>
<mrow>
<mo>(</mo>
<mi>D</mi>
<mi>o</mi>
<mi>G</mi>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mo>|</mo>
<mo>|</mo>
<mi>H</mi>
<mrow>
<mo>(</mo>
<mi>D</mi>
<mi>o</mi>
<mi>G</mi>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>|</mo>
<msub>
<mo>|</mo>
<mn>1</mn>
</msub>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>10</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein,
<mrow>
<mi>D</mi>
<mi>o</mi>
<mi>G</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>2</mn>
<mi>&pi;</mi>
<msup>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mi>&sigma;</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mfrac>
<mrow>
<msup>
<mi>x</mi>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mi>y</mi>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mi>&sigma;</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>2</mn>
<msup>
<mi>&pi;&sigma;</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mfrac>
<mrow>
<msup>
<mi>x</mi>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mi>y</mi>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<mi>&sigma;</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein, | | | |1For (L1) norm, H (x)=max (0, x), DoG (x, y) they are expression formula corresponding to DoG templates;
The inhibition response Inh (x, y) of each pixel is:
Inh (x, y)=Ec (x, y;σ(x,y))*wσ(x,y)σ(x,y) (11)。
6. the profile testing method as claimed in claim 5 based on the global modulation of changeable reception field yardstick, it is characterised in that:
Described step E is specially:
Described profile responds R (x, y):
R (x, y)=H (Ec (x, y;σ(x,y))-αInh(x,y)) (12);
Wherein H (x)=max (0, x), α are inhibition strength.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711098829.XA CN107767387B (en) | 2017-11-09 | 2017-11-09 | Contour detection method based on variable receptive field scale global modulation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711098829.XA CN107767387B (en) | 2017-11-09 | 2017-11-09 | Contour detection method based on variable receptive field scale global modulation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107767387A true CN107767387A (en) | 2018-03-06 |
CN107767387B CN107767387B (en) | 2020-05-05 |
Family
ID=61272282
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711098829.XA Active CN107767387B (en) | 2017-11-09 | 2017-11-09 | Contour detection method based on variable receptive field scale global modulation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107767387B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109766919A (en) * | 2018-12-18 | 2019-05-17 | 通号通信信息集团有限公司 | Cascade the gradual change type Classification Loss calculation method and system in object detection system |
CN109919945A (en) * | 2019-02-01 | 2019-06-21 | 广西科技大学 | Profile testing method based on the non-linear two sides subunit response of non-classical receptive field |
CN109949324A (en) * | 2019-02-01 | 2019-06-28 | 广西科技大学 | Profile testing method based on the non-linear subunit response of non-classical receptive field |
CN109978898A (en) * | 2019-02-01 | 2019-07-05 | 广西科技大学 | Profile testing method based on vector field energy balane |
CN111062957A (en) * | 2019-10-28 | 2020-04-24 | 广西科技大学鹿山学院 | Non-classical receptive field based contour detection method |
CN111080663A (en) * | 2019-12-30 | 2020-04-28 | 广西科技大学 | Bionic contour detection method based on dynamic receptive field |
CN111179293A (en) * | 2019-12-30 | 2020-05-19 | 广西科技大学 | Bionic contour detection method based on color and gray level feature fusion |
CN111539969A (en) * | 2020-04-23 | 2020-08-14 | 武汉铁路职业技术学院 | Image edge detection method and device, computer equipment and storage medium |
CN111968140A (en) * | 2020-06-24 | 2020-11-20 | 广西科技大学 | Contour detection method based on classical receptive field fixation micromotion mechanism |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5422962A (en) * | 1992-03-19 | 1995-06-06 | Fujitsu Limited | Method and apparatus for extracting line segments from an image of an object |
US20020154833A1 (en) * | 2001-03-08 | 2002-10-24 | Christof Koch | Computation of intrinsic perceptual saliency in visual environments, and applications |
US20030118246A1 (en) * | 2001-10-15 | 2003-06-26 | Jonas August | Biased curve indicator random field filters for enhancement of contours in images |
US20080270335A1 (en) * | 2001-05-31 | 2008-10-30 | Canon Kabushiki Kaisha | Pulse signal circuit, parallel processing circuit, and pattern recognition system |
CN101673345A (en) * | 2009-07-01 | 2010-03-17 | 北京交通大学 | Method for extracting target closed contour based on shape prior |
CN102034105A (en) * | 2010-12-16 | 2011-04-27 | 电子科技大学 | Object contour detection method for complex scene |
US20130301910A1 (en) * | 2012-05-14 | 2013-11-14 | University Of Southern California | Extracting object edges from images |
CN103473759A (en) * | 2013-06-24 | 2013-12-25 | 南京理工大学 | Low-light-level image significant contour extraction method of WKPCA homogeneity degree correction nCRF inhibition |
US20140254922A1 (en) * | 2013-03-11 | 2014-09-11 | Microsoft Corporation | Salient Object Detection in Images via Saliency |
CN104484667A (en) * | 2014-12-30 | 2015-04-01 | 华中科技大学 | Contour extraction method based on brightness characteristic and contour integrity |
CN106033609A (en) * | 2015-07-24 | 2016-10-19 | 广西科技大学 | Target contour detection method of biomimetic jumping eye movement information processing mechanism |
CN106485724A (en) * | 2016-09-20 | 2017-03-08 | 华中科技大学 | A kind of profile testing method that modulates based on combination receptive field and towards feature |
-
2017
- 2017-11-09 CN CN201711098829.XA patent/CN107767387B/en active Active
Patent Citations (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5422962A (en) * | 1992-03-19 | 1995-06-06 | Fujitsu Limited | Method and apparatus for extracting line segments from an image of an object |
US20020154833A1 (en) * | 2001-03-08 | 2002-10-24 | Christof Koch | Computation of intrinsic perceptual saliency in visual environments, and applications |
US20080270335A1 (en) * | 2001-05-31 | 2008-10-30 | Canon Kabushiki Kaisha | Pulse signal circuit, parallel processing circuit, and pattern recognition system |
US20030118246A1 (en) * | 2001-10-15 | 2003-06-26 | Jonas August | Biased curve indicator random field filters for enhancement of contours in images |
US7130484B2 (en) * | 2001-10-15 | 2006-10-31 | Jonas August | Biased curve indicator random field filters for enhancement of contours in images |
CN101673345A (en) * | 2009-07-01 | 2010-03-17 | 北京交通大学 | Method for extracting target closed contour based on shape prior |
CN102034105A (en) * | 2010-12-16 | 2011-04-27 | 电子科技大学 | Object contour detection method for complex scene |
US20130301910A1 (en) * | 2012-05-14 | 2013-11-14 | University Of Southern California | Extracting object edges from images |
US8971614B2 (en) * | 2012-05-14 | 2015-03-03 | University Of Southern California | Extracting object edges from images |
US20140254922A1 (en) * | 2013-03-11 | 2014-09-11 | Microsoft Corporation | Salient Object Detection in Images via Saliency |
CN103473759A (en) * | 2013-06-24 | 2013-12-25 | 南京理工大学 | Low-light-level image significant contour extraction method of WKPCA homogeneity degree correction nCRF inhibition |
CN104484667A (en) * | 2014-12-30 | 2015-04-01 | 华中科技大学 | Contour extraction method based on brightness characteristic and contour integrity |
CN106033609A (en) * | 2015-07-24 | 2016-10-19 | 广西科技大学 | Target contour detection method of biomimetic jumping eye movement information processing mechanism |
CN106485724A (en) * | 2016-09-20 | 2017-03-08 | 华中科技大学 | A kind of profile testing method that modulates based on combination receptive field and towards feature |
Non-Patent Citations (6)
Title |
---|
CHUAN LIN 等: "Improved contour detection model with spatial summation properties based on nonclassical receptive field", 《JOURNAL OF ELECTRONIC IMAGING》 * |
COSMIN GRIGORESCU 等: "Contour detection operators based on surround inhibition", 《PROCEEDINGS 2003 INTERNATIONAL CONFERENCE ON IMAGE PROCESSING》 * |
COSMIN GRIGORESCU 等: "Improved Contour Detection by Non-classical Receptive Field Inhibition", 《BMCV 2002》 * |
RONGCHANG ZHAO 等: "Orientation Histogram-Based Center-Surround Interaction: An Integration Approach for Contour Detection", 《NEURAL COMPUTATION》 * |
林川 等: "基于多通道Gabor滤波的指纹图像二值化方法", 《科学技术与工程》 * |
许跃颖 等: "基于非经典感受野多尺度机制的图像分析方法", 《信息技术》 * |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109766919B (en) * | 2018-12-18 | 2020-11-10 | 通号通信信息集团有限公司 | Gradual change type classification loss calculation method and system in cascade target detection system |
CN109766919A (en) * | 2018-12-18 | 2019-05-17 | 通号通信信息集团有限公司 | Cascade the gradual change type Classification Loss calculation method and system in object detection system |
CN109919945A (en) * | 2019-02-01 | 2019-06-21 | 广西科技大学 | Profile testing method based on the non-linear two sides subunit response of non-classical receptive field |
CN109949324A (en) * | 2019-02-01 | 2019-06-28 | 广西科技大学 | Profile testing method based on the non-linear subunit response of non-classical receptive field |
CN109978898A (en) * | 2019-02-01 | 2019-07-05 | 广西科技大学 | Profile testing method based on vector field energy balane |
CN109978898B (en) * | 2019-02-01 | 2023-07-18 | 广西科技大学 | Contour detection method based on vector field energy calculation |
CN109949324B (en) * | 2019-02-01 | 2022-04-22 | 广西科技大学 | Contour detection method based on non-classical receptive field nonlinear subunit response |
CN109919945B (en) * | 2019-02-01 | 2022-03-25 | 广西科技大学 | Contour detection method based on non-classical receptive field non-linear two-side subunit response |
CN111062957A (en) * | 2019-10-28 | 2020-04-24 | 广西科技大学鹿山学院 | Non-classical receptive field based contour detection method |
CN111062957B (en) * | 2019-10-28 | 2024-02-09 | 广西科技大学鹿山学院 | Non-classical receptive field contour detection method |
CN111179293B (en) * | 2019-12-30 | 2020-10-02 | 广西科技大学 | Bionic contour detection method based on color and gray level feature fusion |
CN111080663B (en) * | 2019-12-30 | 2020-09-22 | 广西科技大学 | Bionic contour detection method based on dynamic receptive field |
CN111179293A (en) * | 2019-12-30 | 2020-05-19 | 广西科技大学 | Bionic contour detection method based on color and gray level feature fusion |
CN111080663A (en) * | 2019-12-30 | 2020-04-28 | 广西科技大学 | Bionic contour detection method based on dynamic receptive field |
CN111539969A (en) * | 2020-04-23 | 2020-08-14 | 武汉铁路职业技术学院 | Image edge detection method and device, computer equipment and storage medium |
CN111539969B (en) * | 2020-04-23 | 2023-06-09 | 武汉铁路职业技术学院 | Image edge detection method, device, computer equipment and storage medium |
CN111968140A (en) * | 2020-06-24 | 2020-11-20 | 广西科技大学 | Contour detection method based on classical receptive field fixation micromotion mechanism |
CN111968140B (en) * | 2020-06-24 | 2023-07-14 | 广西科技大学 | Contour detection method based on classical receptive field vision-strengthening micro-motion mechanism |
Also Published As
Publication number | Publication date |
---|---|
CN107767387B (en) | 2020-05-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107767387A (en) | Profile testing method based on the global modulation of changeable reception field yardstick | |
CN104484667B (en) | A kind of contour extraction method based on brightness and integrality of outline | |
CN108022233A (en) | A kind of edge of work extracting method based on modified Canny operators | |
CN106228547B (en) | A kind of profile and border detection algorithm of view-based access control model color theory and homogeneity inhibition | |
CN104299008B (en) | Vehicle type classification method based on multi-feature fusion | |
CN106033610B (en) | Profile testing method based on the modulation of non-classical receptive field spatial summation | |
CN105930848B (en) | SAR image target recognition method based on SAR-SIFT feature | |
CN106033608B (en) | The objective contour detection method of bionical object smooth pursuit eye movement information processing mechanism | |
CN101968850B (en) | Method for extracting face feature by simulating biological vision mechanism | |
CN105354866A (en) | Polygon contour similarity detection method | |
CN107292869B (en) | Image speckle detection method based on anisotropic Gaussian kernel and gradient search | |
CN104834933A (en) | Method and device for detecting salient region of image | |
CN103295014B (en) | Image local feature description method based on pixel location arrangement column diagrams | |
CN105023253A (en) | Visual underlying feature-based image enhancement method | |
CN103473759B (en) | The twilight image method for extracting remarkable configuration that a kind of WKPCA homogeneity degree correction nCRF suppresses | |
CN101763641A (en) | Method for detecting contour of image target object by simulated vision mechanism | |
CN102034224B (en) | Pseudo-Zernike moment-based image denoising algorithm | |
CN108090492A (en) | The profile testing method inhibited based on scale clue | |
CN107886539B (en) | High-precision gear visual detection method in industrial scene | |
CN106780582A (en) | Based on the image significance detection method that textural characteristics and color characteristic are merged | |
CN107067407B (en) | Contour detection method based on non-classical receptive field and linear nonlinear modulation | |
CN107742302A (en) | Profile testing method based on the multiple dimensioned profile fusion of primary visual cortex | |
CN109872326B (en) | Contour detection method based on deep reinforced network jump connection | |
CN104408472B (en) | Classification of Polarimetric SAR Image method based on Wishart and SVM | |
CN107909593A (en) | Non- set direction profile testing method based on receptive field subregion |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20180306 Assignee: GUANGXI YINGTENG EDUCATION TECHNOLOGY Co.,Ltd. Assignor: GUANGXI University OF SCIENCE AND TECHNOLOGY Contract record no.: X2023980053979 Denomination of invention: A contour detection method based on variable receptive field scale global modulation Granted publication date: 20200505 License type: Common License Record date: 20231226 |
|
EE01 | Entry into force of recordation of patent licensing contract |