CN107067407A - Profile testing method based on non-classical receptive field and linear non-linear modulation - Google Patents
Profile testing method based on non-classical receptive field and linear non-linear modulation Download PDFInfo
- Publication number
- CN107067407A CN107067407A CN201710230469.8A CN201710230469A CN107067407A CN 107067407 A CN107067407 A CN 107067407A CN 201710230469 A CN201710230469 A CN 201710230469A CN 107067407 A CN107067407 A CN 107067407A
- Authority
- CN
- China
- Prior art keywords
- mrow
- receptive field
- msub
- central nervous
- classical receptive
- 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
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
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Medicines Containing Antibodies Or Antigens For Use As Internal Diagnostic Agents (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention provides the profile testing method based on non-classical receptive field and linear non-linear modulation, including：A, the image to be detected of input through gray proces；B, the Gabor energy values that all directions for obtaining each pixel are filtered to image to be detected progress Gabor；C, the X cell and Y cell simulation model built in retinal ganglial cells；D, calculating obtain stimuli responsive of corresponding each central nervous member of X cell by non-classical receptive field；E, calculating obtain stimuli responsive of corresponding each central nervous member of Y cell by non-classical receptive field；F, the central nervous member for calculating X, Y cell respectively are used as corresponding profile value by classical receptive field and the stimuli responsive of non-classical receptive field combined modulation after addition；G, the profile value progress to each pixel handle and obtain final profile value.This method overcomes the low defect of prior art outline identification rate, with the characteristics of simulated effect is good, outline identification rate is high.
Description
Technical field
The present invention relates to image processing field, and in particular to a kind of based on non-classical receptive field and linear non-linear modulation
Profile testing method.
Background technology
The shape of outline definition target, profile is one of vital task in target identification, and is obtained from mixed and disorderly scene
Objective contour be an important and extremely difficult task, be primarily due to around profile generally the presence of a large amount of grain backgrounds
Edge, therefore the main needs exclusion of this work is due to the meaningless edge of texture region, and retain objective contour.Improve detection
The key of rate is local message can be optimized and combined into consistent global characteristics based on context.Human visual system has fast
Speed and the effective ability that contour feature is extracted from complex scene, have effectively facilitated the profile inspection using biological nature as inspiration
The development of method of determining and calculating research.Physiological Study shows that V1 layers of neuron have orientation selectivity, and in its classical receptive field
There is non-classical receptive field (Non-Classical Receptive outside (Classical Receptive Rield, CRF)
Rield, NCRF) region, although individually stimulate the region not respond to, but there can be certain modulating action to CRF.
Rodieck proposes the mathematical modeling on concentric circles antagonism formula receptive field in nineteen sixty-five, and it is by an excited work
Constituted with strong center mechanism and the weaker but bigger area inhibition periphery mechanism of effect.Rodieck models are also known as
Gaussian difference model, difference of two Gaussians, DOG.Enroth-Cugell in 1966 and Robson observation cats
GC can be divided into two classes by the space-time summation property of its reaction：The spatial summation characteristic of one class cell effect is substantially
X cell can be claimed with linear, additive by meeting Rodieck models, i.e. their receptive field excitement and inhibitory action；Another kind of GC
Spatial summation property be nonlinear, Rodieck models are not very applicable it, referred to as Y cell.Enroth-Cugell et
Al. find that the receptive field spatial character of some cells (X cell) in gangliocyte is similar to linear, and other cells (Y
Cell) nonlinear space characteristics are largely shown, these characteristics are also existed in foreign journals cell.
The content of the invention
The present invention is intended to provide a kind of profile testing method based on non-classical receptive field and linear non-linear modulation, the party
Method overcomes the defect that prior art simulated effect is poor, outline identification rate is low, with the spy that simulated effect is good, outline identification rate is high
Point.
Technical scheme is as follows：Profile testing method based on non-classical receptive field and linear non-linear modulation,
Comprise the following steps：
A, the image to be detected of input through gray proces, using each pixel of image to be detected as non-classical sense
By Yezhong cardiac nerve member；
B, default multiple directions parameter Gabor filter group, to each pixel in image to be detected respectively according to each
Individual directioin parameter carries out Gabor filtering, obtains the Gabor energy values of all directions of each pixel；For each pixel, choosing
The maximum in the Gabor energy values of its all directions is taken, as stimuli responsive of the pixel by classical receptive field, as should
Non-classical receptive field central nervous member stimuli responsive, the corresponding filtering direction of the maximum as the pixel optimal corner,
As the non-classical receptive field central nervous member optimal corner；
C, the X cell and Y cell simulation model built in retinal ganglial cells；
D, the response for X cell path are calculated：
Based on each non-classical receptive field central nervous member, each non-classical receptive field is calculated to its central nervous member
Spatial summation modulates weights, and is normalized, while calculating distance of each non-classical receptive field to its central nervous member
Weights respond, then by each non-classical receptive field to its central nervous member apart from weights response with it is corresponding normalized
Spatial summation modulation weights, which are multiplied, obtains the first stimuli responsive by non-classical receptive field of each corresponding central nervous of X cell；
E, the response for Y cell path are calculated：
Multiple directions are uniformly arranged along non-classical receptive field central nervous member outer interval, it is pre- in default all directions
If multiple basic points, the basic point for calculating each non-classical receptive field modulates weights to the spatial summation of its central nervous member, and will
It is normalized, while calculating basic point being responded apart from weights to its central nervous member of each non-classical receptive field, is passed through
The basic point of each non-classical receptive field is adjusted to its central nervous member apart from weights response with corresponding normalized spatial summation
Weights processed, which are multiplied, obtains the first stimuli responsive by non-classical receptive field of each corresponding central nervous of Y cell；
F, X cell, Y cell corresponding each central nervous member are obtained respectively joined by classical receptive field and non-classical receptive field
The stimuli responsive of modulation is closed, above-mentioned X, the corresponding combined modulation of Y cell stimuli responsive are taken and are just added afterwards, additive value is used as this
The profile value of central nervous member corresponding pixel points；
G, the profile value to each pixel are handled using non-maxima suppression and dual threshold, obtain the final wheel of each pixel
Exterior feature value.
Preferably, in described step D：
The contrast weighting function for obtaining each non-classical receptive field to its central nervous member, and basis respectively are calculated respectively
The corresponding optimal corner of neuron, which is calculated, in each non-classical receptive field obtains each non-classical receptive field central nervous member towards weight letter
Number, and then spatial summation modulation weights of each non-classical receptive field to its central nervous member are calculated, and normalized；
Distance weighting function of each non-classical receptive field to its central nervous member is calculated, by this distance weighting function
Convolution is carried out by the stimuli responsive of classical receptive field obtain each non-classical receptive field to its central nervous with its central nervous member
First responds apart from weights.
Preferably, in described step E：
Described basic point prolongs default directional spreding in non-classical receptive field, and Gauss weighting is carried out to each basic point and obtains each base
The energy and angle of point；The contrast weighting function for obtaining each basic point to its central nervous member is calculated respectively, and respectively according to each
The angle calculation of basic point obtains each non-classical receptive field central nervous member towards weighting function, and then calculates each non-classical sense
Weights are modulated to the spatial summation of its central nervous member by wild basic point, and normalized；
Distance weighting function of the basic point to its central nervous member of each non-classical receptive field is calculated, this distance is weighed
Weight function is multiplied the basic point for obtaining each non-classical receptive field to it with its central nervous member by the stimuli responsive of classical receptive field
Central nervous is first to be responded apart from weights；
By the basic point of each non-classical receptive field to its central nervous member apart from weights response and corresponding normalization
Spatial summation modulation weights be multiplied the stimuli responsive for obtaining Y cell corresponding each central nervous member by non-classical receptive field.
Preferably, in described step F, respectively by X cell, corresponding each central nervous member of Y cell by non-classical sense
Stimuli responsive by open country modulation is multiplied with rejection coefficient, then its central nervous member is subtracted by the stimuli responsive of classical receptive field
State the stimuli responsive that product respectively obtains the corresponding combined modulation of X cell, Y cell.
Preferably, described step B is specific as follows：
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 is the bandwidth in area of DoG template center, and 1/ λ is the spatial frequency of cosine function, and σ/λ is space frequency
The bandwidth of rate,It is phase angular dimensions, θ is the angle parameter that Gabor is filtered；
I (x, y) is image to be detected, and * is convolution operator；
Gabor energy values are calculated as follows：
Wherein θiThe a certain angle filtered for Gabor, NθFor the number of the Gabor angles filtered；
Ec (x, y) is the maximum of each angle Gabor filtered energy values of pixel (x, y),It is right for Ec (x, y)
The filtering angle answered, is used as the optimal corner of pixel (x, y).
Preferably, described step D is specific as follows：
For X cell：
Described non-classical receptive field to central nervous member spatial summation modulate weights expression formula be：
inhgox(x, y)=∑x′∑y′wgx(x,y)wox(x+x′,y+y′) (7)；
Wherein -3k σ<x′<3kσ；-3kσ<y′<3kσ；
Wherein inhgox(x, y) is that non-classical receptive field modulates weights, w to the spatial summation of central nervous membergx(x, y) is
Non-classical receptive field is to the contrast weighting function of central nervous member, wox(x+x ', y+y ') is the non-classical receptive field centering mind
Through first towards weighting function；
wgx(x, y) expression formula is：
Wherein, EAVGFor average of each pixel of image to be detected by classical receptive field stimuli responsive, as image to be detected
The average of Ec (x, y) value of each pixel；
woxThe expression formula of (x+x ', y+y ') is：
Wherein ω=max (ω1, ω2)；
WhereinCentered on neuron A (x, y) optimal corner,For appointing in addition to central nervous member in non-classical receptive field
The corresponding filtering angle of maximum gabor energy values of one neuron B (x+x ', y+y ') optimal corner, i.e. respectively neuron A, B
Degree,Centered on neuron A and neuron B lines the deviation angle；
Weights are modulated for normalized spatial summation；
Described non-classical receptive field is expressed as to central nervous member apart from weights response：
inhdx(x, y)=Ec (x, y) wd(x,y) (10)；
Wherein inhdx(x, y) is non-classical receptive field being responded apart from weights to central nervous member, wd(x, y) is non-classical
Distance weighting function of the receptive field to central nervous member；
Central nervous member is by the stimuli responsive expression formula of non-classical receptive field：
Preferably, described step E is specific as follows：
For Y cell：
The coordinate of each basic point is (Δ xi,j,Δyi,j), Δ xi,j=dicosφj、Δyi,j=disinφj；
Wherein dI=1,2...5=(2 σ, 4 σ, 6 σ, 8 σ, 10 σ), φJ=1,2 ... 8=(0, π/4, pi/2,3 π/4, π, 5/4,3 pi/2,
7π/4)；
The energy and angle calculation of each basic point are as follows：
Wherein Gσ(x′,y′；σG) it is Gaussian function, σGThe σ of=σ, x '≤3G, the σ of y '≤3G；T is the multiplier of scale parameter；
Each basic point is to the expression formula of the spatial summation modulation weights of central nervous member in described non-classical receptive field：
inhgoy(x, y)=∑x′∑y′wgy(x,y)woy(x+x′,y+y′) (14)；
Wherein -3k σ<x′<3kσ；-3kσ<y′<3kσ；
Wherein inhgoy(x, y) modulates weights, w for each basic point in non-classical receptive field to the spatial summation of central nervous membergy
(x, y) is contrast weighting function of the non-classical receptive field to central nervous member, woy(x+x ', y+y ') is non-classical receptive field pair
Central nervous member towards weighting function；
wgy(x, y) expression formula is：
Wherein, EAVGFor average of each pixel of image to be detected by classical receptive field stimuli responsive, as image to be detected
The average of Ec (x, y) value of each pixel；
woyThe expression formula of (x, y) is：
Wherein ω=max (ω1, ω2)；
WhereinCentered on neuron A (x, y) optimal corner,For appointing in addition to central nervous member in non-classical receptive field
One basic point B angle,Centered on neuron A and basic point neuron B lines the deviation angle；
Weights are modulated for normalized spatial summation；
Described non-classical receptive field is expressed as to central nervous member apart from weights response：
Wherein inhdy(x, y) is basic point being responded apart from weights to central nervous member of non-classical receptive field, wd(x, y) is
Distance weighting function of the non-classical receptive field to central nervous member；
Wherein, wdThe expression formula of (x, y) is：
Wherein,
Wherein, | | | |1For (L1) norm, H (DoG (x, y)) be take on the occasion of function, DoG (x, y) be DoG templates correspondingly
Expression formula；
Central nervous member is by the stimuli responsive expression formula of non-classical receptive field：
Preferably, described step F is specific as follows：
For X cell：
Expression formula of the described central nervous member by classical receptive field and the stimuli responsive of non-classical receptive field combined modulation
For：
Rx(x, y)=H (Ec (x, y)-α inhx(x,y)) (21)；
Wherein Rx(x, y) is the corresponding central nervous member of X cell by classical receptive field and non-classical receptive field combined modulation
Stimuli responsive, α be the corresponding rejection coefficient of X cell；
For Y cell：
Expression formula of the described central nervous member by classical receptive field and the stimuli responsive of non-classical receptive field combined modulation
For：
Ry(x, y)=H (Ec (x, y)-β inhy(x,y)) (22)；
Wherein Ry(x, y) is the corresponding central nervous member of Y cell by classical receptive field and non-classical receptive field combined modulation
Stimuli responsive, β be the corresponding rejection coefficient of Y cell；
The described positive function that takes is
The profile value expression of the described corresponding pixel of central nervous member is as follows：
R (x, y)=Rx(x,y)+Ry(x,y) (23)；
The profile value of the corresponding pixel of neuron centered on wherein R (x, y).
The present invention is simulated except the impression response to X cell, while also bigger than X cell based on Y cell receptive field
Feature, the impression response with the multiplier of scale parameter to Y cell is emulated；Innovatively by setting in multiple directions
Basic point, Gauss weighting is carried out to basic point roundlet region and is integrated, so as to realize the analog simulation of Y cell impression response, basic point is set
Fixed pretreatment, can more preserve stable profile information；The area information around basic point is integrated by Gauss weighting, both
It can realize that Y cell experiences the emulation of response property, the amount of calculation of high yardstick can be effectively reduced again, improve operation efficiency.
Brief description of the drawings
Fig. 1 is profile testing method flow chart of the invention
Fig. 2 is the method for embodiment 1 and the Detection results comparison diagram of the contour detecting model of file 1
Fig. 3 is the method for embodiment 1 and the detection comparative bid parameter of the contour detecting model of file 1
Fig. 4 is the basic point schematic diagram of the profile testing method of the present invention
Embodiment
Embodiment 1
The profile testing method based on non-classical receptive field and linear non-linear modulation that the present embodiment is provided, including it is following
Step：
A, the image to be detected of input through gray proces, using each pixel of image to be detected as non-classical sense
By Yezhong cardiac nerve member；
B, default multiple directions parameter Gabor filter group, to each pixel in image to be detected respectively according to each
Individual directioin parameter carries out Gabor filtering, obtains the Gabor energy values of all directions of each pixel；For each pixel, choosing
The maximum in the Gabor energy values of its all directions is taken, as stimuli responsive of the pixel by classical receptive field, as should
Non-classical receptive field central nervous member stimuli responsive, the corresponding filtering direction of the maximum as the pixel optimal corner,
As the non-classical receptive field central nervous member optimal corner；
Described step B is specific as follows：
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 is the bandwidth in area of DoG template center, and 1/ λ is the spatial frequency of cosine function, and σ/λ is space frequency
The bandwidth of rate,It is phase angular dimensions, θ is the angle parameter that Gabor is filtered；
I (x, y) is image to be detected, and * is convolution operator；
Gabor energy values are calculated as follows：
Wherein θiThe a certain angle filtered for Gabor, NθFor the number of the Gabor angles filtered；
Ec (x, y) is the maximum of each angle Gabor filtered energy values of pixel (x, y),It is right for Ec (x, y)
The filtering angle answered, is used as the optimal corner of pixel (x, y)；
C, the X cell and Y cell simulation model built in retinal ganglial cells；
D, the response for X cell path are calculated：
Based on each non-classical receptive field central nervous member, calculate obtain each non-classical receptive field to the wherein mind respectively
Contrast weighting function through member, and obtain each non-according to the corresponding optimal corner calculating of neuron in each non-classical receptive field respectively
Classical receptive field central nervous member calculates sky of each non-classical receptive field to its central nervous member towards weighting function
Between summation modulation weights, and normalized；Each non-classical receptive field is calculated simultaneously to weigh the distance of its central nervous member
Weight function, this distance weighting function is multiplied with its central nervous member by the stimuli responsive of classical receptive field and obtains each non-warp
Allusion quotation receptive field is responded to its central nervous member apart from weights；Then by each non-classical receptive field to its central nervous member
The corresponding mind in each of X cell is obtained apart from weights response and corresponding normalized spatial summation modulation weights progress convolution
Through stimuli responsive of the member by non-classical receptive field；
For X cell：
Described non-classical receptive field to central nervous member spatial summation modulate weights expression formula be：
inhgox(x, y)=∑x′∑y′wgx(x,y)wox(x+x′,y+y′) (7)；
Wherein -3k σ<x′<3kσ；-3kσ<y′<3kσ；
Wherein inhgox(x, y) is that non-classical receptive field modulates weights, w to the spatial summation of central nervous membergx(x, y) is
Non-classical receptive field is to the contrast weighting function of central nervous member, wox(x+x ', y+y ') is the non-classical receptive field centering mind
Through first towards weighting function；
wgx(x, y) expression formula is：
Wherein, EAVGFor average of each pixel of image to be detected by classical receptive field stimuli responsive, as image to be detected
The average of Ec (x, y) value of each pixel；
woxThe expression formula of (x+x ', y+y ') is：
Wherein ω=max (ω1, ω2)；
WhereinCentered on neuron A (x, y) optimal corner,For appointing in addition to central nervous member in non-classical receptive field
The corresponding filtering angle of maximum gabor energy values of one neuron B (x+x ', y+y ') optimal corner, i.e. respectively neuron A, B
Degree,Centered on neuron A and neuron B lines the deviation angle；
Weights are modulated for normalized spatial summation；
Described non-classical receptive field is expressed as to central nervous member apart from weights response：
inhdx(x, y)=Ec (x, y) wd(x,y) (10)；
Wherein inhdx(x, y) is non-classical receptive field being responded apart from weights to central nervous member, wd(x, y) is non-classical
Distance weighting function of the receptive field to central nervous member；
Central nervous member is by the stimuli responsive expression formula of non-classical receptive field：
E, the response for Y cell path are calculated：
Multiple directions are uniformly arranged along non-classical receptive field central nervous member outer interval, basic point prolongs default directional spreding
In non-classical receptive field, energy and angle that Gauss weighting obtains each basic point are carried out to each basic point；Calculate respectively and obtain each basic point
To the contrast weighting function of its central nervous member, and obtained respectively according to the angle calculation of each basic point in each non-classical receptive field
Cardiac nerve member calculates spatial summation of the basic point to its central nervous member of each non-classical receptive field towards weighting function
Weights are modulated, and are normalized；
Distance weighting function of the basic point to its central nervous member of each non-classical receptive field is calculated, this distance is weighed
Weight function is multiplied the basic point for obtaining each non-classical receptive field to it with its central nervous member by the stimuli responsive of classical receptive field
Central nervous is first to be responded apart from weights；
By the basic point of each non-classical receptive field to its central nervous member apart from weights response and corresponding normalization
Spatial summation modulation weights carry out convolution and obtain Y cell corresponding each central nervous member being rung by the stimulation of non-classical receptive field
Should；
The coordinate of each basic point is (Δ xi,j,Δyi,j), Δ xi,j=dicosφj、Δyi,j=disinφj；
Wherein dI=1,2...5=(2 σ, 4 σ, 6 σ, 8 σ, 10 σ), φJ=1,2 ... 8=(0, π/4, pi/2,3 π/4, π, 5/4,3 pi/2,
7π/4)；
The energy and angle calculation of each basic point are as follows：
Wherein Gσ(x′,y′；σG) it is Gaussian function, σGThe σ of=σ, x '≤3G, the σ of y '≤3G；T is the multiplier of scale parameter；
Each basic point is to the expression formula of the spatial summation modulation weights of central nervous member in described non-classical receptive field：
inhgoy(x, y)=∑x′∑y′wgy(x,y)woy(x+x′,y+y′) (14)；
Wherein -3k σ<x′<3kσ；-3kσ<y′<3kσ；
Wherein inhgoy(x, y) modulates weights, w for each basic point in non-classical receptive field to the spatial summation of central nervous membergy
(x, y) is contrast weighting function of the non-classical receptive field to central nervous member, woy(x+x ', y+y ') is non-classical receptive field pair
Central nervous member towards weighting function；
wgy(x, y) expression formula is：
Wherein, EAVGFor average of each pixel of image to be detected by classical receptive field stimuli responsive, as image to be detected
The average of Ec (x, y) value of each pixel；
woyThe expression formula of (x, y) is：
Wherein ω=max (ω1, ω2)；
WhereinCentered on neuron A (x, y) optimal corner,For appointing in addition to central nervous member in non-classical receptive field
One basic point B angle,Centered on neuron A and basic point neuron B lines the deviation angle；
Weights are modulated for normalized spatial summation；
Described non-classical receptive field is expressed as to central nervous member apart from weights response：
Wherein inhdy(x, y) is basic point being responded apart from weights to central nervous member of non-classical receptive field, wd(x, y) is
Distance weighting function of the non-classical receptive field to central nervous member；
Wherein, wdThe expression formula of (x, y) is：
Wherein,
Wherein, | | | |1For (L1) norm, H (DoG (x, y)) be take on the occasion of function, DoG (x, y) be DoG templates correspondingly
Expression formula；
Central nervous member is by the stimuli responsive expression formula of non-classical receptive field：
F, the stimuli responsive for respectively being modulated corresponding each central nervous member of X cell, Y cell by non-classical receptive field with
Rejection coefficient be multiplied, then by its central nervous member by the stimuli responsive of classical receptive field subtract above-mentioned product respectively obtain X cell,
The stimuli responsive of the corresponding combined modulation of Y cell, phase after above-mentioned X, the corresponding combined modulation of Y cell stimuli responsive are taken just
Plus, additive value as the central nervous member corresponding pixel points profile value；
For X cell：
Expression formula of the described central nervous member by classical receptive field and the stimuli responsive of non-classical receptive field combined modulation
For：
Rx(x, y)=H (Ec (x, y)-α inhx(x,y)) (21)；
Wherein Rx(x, y) is the corresponding central nervous member of X cell by classical receptive field and non-classical receptive field combined modulation
Stimuli responsive, α be the corresponding rejection coefficient of X cell；
For Y cell：
Expression formula of the described central nervous member by classical receptive field and the stimuli responsive of non-classical receptive field combined modulation
For：
Ry(x, y)=H (Ec (x, y)-β inhy(x,y)) (22)；
Wherein Ry(x, y) is the corresponding central nervous member of Y cell by classical receptive field and non-classical receptive field combined modulation
Stimuli responsive, β be the corresponding rejection coefficient of X cell；
The described positive function that takes is
The profile value expression of the described corresponding pixel of central nervous member is as follows：
R (x, y)=Rx(x,y)+Ry(x,y) (23)；
The profile value of the corresponding pixel of neuron centered on wherein R (x, y).
The contour detecting isotropic model and items that the profile testing method of the present embodiment and document 1 are provided below
Different in nature model carries out Usefulness Pair ratio, wherein being carried out from the isotropic model and anisotropic model in document 1 effective
Property contrast, document 1 is as follows：
Document 1：Grigorescu C,Petkov N,Westenberg M.Contour detection based on
nonclassical receptive field inhibition[J].IEEE Transactions on Image
Processing,2003,12(7):729-739；
To ensure the validity of contrast, use to enter with identical non-maxima suppression method in document 1 for the present embodiment
The follow-up profile of row is integrated, wherein the two threshold value t includedh,tlIt is set to tl=0.5th, calculated by threshold value quantile p and obtained；
Wherein Performance Evaluating Indexes P uses the following standard provided in document 1：
N in formulaTP、nFP、nFNThe number of the profile of correct profile, error profile and omission that detection is obtained is represented respectively,
Evaluation metricses P values represent that the effect of contour detecting is better between [0,1] closer to 1, in addition, definition tolerance is：In 5*
The all calculations detected in 5 neighborhood are correctly detected.
Choose the secondary classic map picture of rhinoceros, lion, elephant 3 and carry out Usefulness Pair ratio, the isotropic in document 1 is respectively adopted
Model, anisotropic model and the method for embodiment 1 carry out contour detecting, the wherein method selection of embodiment 1 to above-mentioned 3 width figure
Parameter group is as shown in table 1,
The parameter group table of 1 embodiment of table 1
Isotropic model, anisotropic model in document 1 use 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 }；
One group of best parameter of effect is carried out in selection isotropic model, anisotropic model and the method for embodiment 1
Contrast, the contrast of contours extract design sketch are as shown in Fig. 2 corresponding performance indications contrast is as shown in Figure 3；It can be seen by Fig. 2, Fig. 3
Go out, no matter from the effect of contours extract or from performance indications parameter, the method for embodiment 1 is superior to each in document 1
Item same sex model, anisotropic model；Fig. 4 is the schematic diagram of the Y cell receptive field basic point of embodiment 1, and stain therein is base
Point；
Wherein table 2 is the corresponding partial parameters table of the result figure of embodiment 1, and remaining parameter is with reference to the data in table 1；Table 3,4
Respectively isotropic model, the corresponding parameter list of anisotropic model result figure.
The corresponding partial parameters table of the result figure of 2 embodiment of table 1
The corresponding parameter list of the isotropic model result figure of table 3
The corresponding parameter list of the anisotropic model result figure of table 4
Claims (8)
1. the profile testing method based on non-classical receptive field and linear non-linear modulation, it is characterised in that comprise the following steps：
A, the image to be detected of input through gray proces, using each pixel of image to be detected as non-classical receptive field
Central nervous member；
B, default multiple directions parameter Gabor filter group, to each pixel in image to be detected respectively according to each side
Gabor filtering is carried out to parameter, the Gabor energy values of all directions of each pixel are obtained；For each pixel, it is chosen
Maximum in the Gabor energy values of all directions, is the non-warp as stimuli responsive of the pixel by classical receptive field
The stimuli responsive of allusion quotation receptive field central nervous member, the corresponding filtering direction of the maximum is as the optimal corner of the pixel
The optimal corner of non-classical receptive field central nervous member；
C, the X cell and Y cell simulation model built in retinal ganglial cells；
D, the response for X cell path are calculated：
Based on each non-classical receptive field central nervous member, space of each non-classical receptive field to its central nervous member is calculated
Summation modulates weights, and is normalized, while calculating each non-classical receptive field to its central nervous member apart from weights
Response, is then responded and corresponding normalized space by the way that each non-classical receptive field is first to its central nervous apart from weights
Summation modulation weights, which are multiplied, obtains the first stimuli responsive by non-classical receptive field of each corresponding central nervous of X cell；
E, the response for Y cell path are calculated：
Multiple directions are uniformly arranged along non-classical receptive field central nervous member outer interval, are preset in default all directions many
Individual basic point, the basic point for calculating each non-classical receptive field modulates weights to the spatial summation of its central nervous member, and is returned
One changes, while basic point being responded apart from weights to its central nervous member of each non-classical receptive field is calculated, by each
The basic point of non-classical receptive field is weighed to its central nervous member apart from weights response with corresponding normalized spatial summation modulation
Value, which is multiplied, obtains the first stimuli responsive by non-classical receptive field of each corresponding central nervous of Y cell；
F, X cell, Y cell corresponding each central nervous member are obtained respectively adjusted by classical receptive field and non-classical receptive field joint
The stimuli responsive of system, above-mentioned X, the corresponding combined modulation of Y cell stimuli responsive are taken and are just added afterwards, additive value is used as the center
The profile value of neuron corresponding pixel points；
G, the profile value to each pixel are handled using non-maxima suppression and dual threshold, obtain the final profile of each pixel
Value.
2. the profile testing method as claimed in claim 1 based on non-classical receptive field and linear non-linear modulation, its feature
It is：
In described step D：
The contrast weighting function for obtaining each non-classical receptive field to its central nervous member is calculated respectively, and respectively according to each non-
The corresponding optimal corner of neuron, which is calculated, in classical receptive field obtains each non-classical receptive field central nervous member towards weighting function, enters
And calculate each non-classical receptive field and weights are modulated to the spatial summation of its central nervous member, and normalized；
Distance weighting function of each non-classical receptive field to its central nervous member is calculated, by this distance weighting function and its
Central nervous member is carried out convolution by the stimuli responsive of classical receptive field and obtains each non-classical receptive field to its central nervous member
Apart from weights response.
3. the profile testing method as claimed in claim 2 based on non-classical receptive field and linear non-linear modulation, its feature
It is：
In described step E：
Described basic point prolongs default directional spreding in non-classical receptive field, and Gauss weighting is carried out to each basic point and obtains each basic point
Energy and angle；The contrast weighting function for obtaining each basic point to its central nervous member is calculated respectively, and respectively according to each basic point
Angle calculation obtain each non-classical receptive field central nervous member towards weighting function, and then calculate each non-classical receptive field
Basic point weights are modulated to the spatial summation of its central nervous member, and normalized；
Distance weighting function of the basic point to its central nervous member of each non-classical receptive field is calculated, by this distance weighting letter
Number is carried out convolution by the stimuli responsive of classical receptive field with its central nervous member and obtains the basic point of each non-classical receptive field to it
Central nervous is first to be responded apart from weights；
By the basic point of each non-classical receptive field to its central nervous member apart from weights response and corresponding normalized sky
Between summation modulation weights carry out convolution and obtain stimuli responsive of Y cell corresponding each central nervous member by non-classical receptive field.
4. the profile testing method as claimed in claim 3 based on non-classical receptive field and linear non-linear modulation, its feature
It is：
In described step F, the thorn for respectively being modulated X cell, corresponding each central nervous member of Y cell by non-classical receptive field
Swash response to be multiplied with rejection coefficient, then its central nervous member is subtracted into above-mentioned product by the stimuli responsive of classical receptive field and obtain respectively
To X cell, the stimuli responsive of the corresponding combined modulation of Y cell.
5. the profile testing method as claimed in claim 4 based on non-classical receptive field and linear non-linear modulation, its feature
It is：
Described step B is specific as follows：
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 is wavelength, and σ is the bandwidth in area of DoG template center, and 1/ λ is the spatial frequency of cosine function, and σ/λ is spatial frequency
Bandwidth,It is phase angular dimensions, θ is the angle parameter that Gabor is filtered；
I (x, y) is 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>
<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>
<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>
<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>3</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>4</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein θiThe a certain angle filtered for Gabor, NθFor the number of the Gabor angles filtered；
<mrow>
<mi>E</mi>
<mi>c</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<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>
<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>5</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Ec (x, y) is the maximum of each angle Gabor filtered energy values of pixel (x, y),It is corresponding for Ec (x, y)
Angle is filtered, the optimal corner of pixel (x, y) is used as.
6. the profile testing method as claimed in claim 5 based on non-classical receptive field and linear non-linear modulation, its feature
It is：
Described step D is specific as follows：
For X cell：
Described non-classical receptive field to central nervous member spatial summation modulate weights expression formula be：
inhgox(x, y)=∑x′∑y′wgx(x,y)wox(x+x′,y+y′) (7)；
Wherein -3k σ<x′<3kσ；-3kσ<y′<3kσ；
Wherein inhgox(x, y) is that non-classical receptive field modulates weights, w to the spatial summation of central nervous membergx(x, y) is non-warp
Allusion quotation receptive field is to the contrast weighting function of central nervous member, wox(x+x ', y+y ') is non-classical receptive field to central nervous member
Towards weighting function；
wgx(x, y) expression formula is：
<mrow>
<msub>
<mi>w</mi>
<mrow>
<mi>g</mi>
<mi>x</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>1</mn>
<mo>+</mo>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mi>E</mi>
<mi>c</mi>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>E</mi>
<mrow>
<mi>A</mi>
<mi>V</mi>
<mi>G</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein, EAVGFor average of each pixel of image to be detected by classical receptive field stimuli responsive, as image to be detected each
The average of Ec (x, y) value of pixel；
woxThe expression formula of (x+x ', y+y ') is：
<mrow>
<msub>
<mi>w</mi>
<mrow>
<mi>o</mi>
<mi>x</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>+</mo>
<msup>
<mi>x</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>y</mi>
<mo>+</mo>
<msup>
<mi>y</mi>
<mo>&prime;</mo>
</msup>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>2</mn>
<mi>&pi;</mi>
</mfrac>
<mo>&CenterDot;</mo>
<mi>&omega;</mi>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>9</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein ω=max (ω1, ω2)；
WhereinCentered on neuron A (x, y) optimal corner,For any god in non-classical receptive field in addition to central nervous member
The corresponding filtering angle of maximum gabor energy values of optimal corner through first B (x+x ', y+y '), i.e. respectively neuron A, B,Centered on neuron A and neuron B lines the deviation angle；
Weights are modulated for normalized spatial summation；
Described non-classical receptive field is expressed as to central nervous member apart from weights response：
inhdx(x, y)=Ec (x, y) wd(x,y) (10)；
Wherein inhdx(x, y) is non-classical receptive field being responded apart from weights to central nervous member, wd(x, y) is non-classical impression
The wild distance weighting function to central nervous member；
Central nervous member is by the stimuli responsive expression formula of non-classical receptive field：
7. the profile testing method as claimed in claim 6 based on non-classical receptive field and linear non-linear modulation, its feature
It is：
Described step E is specific as follows：
For Y cell：
The coordinate of each basic point is (Δ xi,j,Δyi,j), Δ xi,j=dicosφj、Δyi,j=disinφj；
Wherein dI=1,2...5=(2 σ, 4 σ, 6 σ, 8 σ, 10 σ), φJ=1,2 ... 8=(0, π/4, pi/2,3 π/4, π, 5/4,3 pi/2,7 π/
4)；
The energy and angle calculation of each basic point are as follows：
<mrow>
<msubsup>
<mi>r</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mi>E</mi>
</msubsup>
<mo>=</mo>
<msub>
<mi>&Sigma;</mi>
<msup>
<mi>x</mi>
<mo>&prime;</mo>
</msup>
</msub>
<msub>
<mi>&Sigma;</mi>
<msup>
<mi>y</mi>
<mo>&prime;</mo>
</msup>
</msub>
<mi>E</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>-</mo>
<msub>
<mi>&Delta;x</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>-</mo>
<msup>
<mi>x</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>y</mi>
<mo>-</mo>
<msub>
<mi>&Delta;y</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>-</mo>
<msup>
<mi>y</mi>
<mo>&prime;</mo>
</msup>
<mo>;</mo>
<mi>T</mi>
<mi>&sigma;</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>G</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<msup>
<mi>x</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<msup>
<mi>y</mi>
<mo>&prime;</mo>
</msup>
<mo>;</mo>
<msub>
<mi>&sigma;</mi>
<mi>G</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>12</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
<mrow>
<msubsup>
<mi>r</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mi>&theta;</mi>
</msubsup>
<mo>=</mo>
<msub>
<mi>&Sigma;</mi>
<msup>
<mi>x</mi>
<mo>&prime;</mo>
</msup>
</msub>
<msub>
<mi>&Sigma;</mi>
<msup>
<mi>y</mi>
<mo>&prime;</mo>
</msup>
</msub>
<mi>&theta;</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>-</mo>
<msub>
<mi>&Delta;x</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>-</mo>
<msup>
<mi>x</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>y</mi>
<mo>-</mo>
<msub>
<mi>&Delta;y</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>-</mo>
<msup>
<mi>y</mi>
<mo>&prime;</mo>
</msup>
<mo>;</mo>
<mi>T</mi>
<mi>&sigma;</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>G</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<msup>
<mi>x</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<msup>
<mi>y</mi>
<mo>&prime;</mo>
</msup>
<mo>;</mo>
<msub>
<mi>&sigma;</mi>
<mi>G</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>13</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein Gσ(x′,y′；σG) it is Gaussian function, σGThe σ of=σ, x '≤3G, the σ of y '≤3G；T is the multiplier of scale parameter；
Each basic point is to the expression formula of the spatial summation modulation weights of central nervous member in described non-classical receptive field：
inhgoy(x, y)=∑x′∑y′wgy(x,y)woy(x+x′,y+y′) (14)；
Wherein -3k σ<x′<3kσ；-3kσ<y′<3kσ；
Wherein inhgoy(x, y) modulates weights, w for each basic point in non-classical receptive field to the spatial summation of central nervous membergy(x,
Y) it is contrast weighting function of the non-classical receptive field to central nervous member, woy(x+x ', y+y ') is non-classical receptive field centering
Cardiac nerve member towards weighting function；
wgy(x, y) expression formula is：
<mrow>
<msub>
<mi>w</mi>
<mrow>
<mi>g</mi>
<mi>y</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>1</mn>
<mo>+</mo>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<msubsup>
<mi>r</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mi>E</mi>
</msubsup>
<mo>+</mo>
<msub>
<mi>E</mi>
<mrow>
<mi>A</mi>
<mi>V</mi>
<mi>G</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>15</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein, EAVGFor average of each pixel of image to be detected by classical receptive field stimuli responsive, as image to be detected each
The average of Ec (x, y) value of pixel；
woyThe expression formula of (x, y) is：
<mrow>
<msub>
<mi>w</mi>
<mrow>
<mi>o</mi>
<mi>y</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>2</mn>
<mi>&pi;</mi>
</mfrac>
<mo>&CenterDot;</mo>
<mi>&omega;</mi>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>16</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein ω=max (ω1, ω2)；
WhereinCentered on neuron A (x, y) optimal corner,For any base in non-classical receptive field in addition to central nervous member
Point B angle,Centered on neuron A and basic point neuron B lines the deviation angle；
Weights are modulated for normalized spatial summation；
Described non-classical receptive field is expressed as to central nervous member apart from weights response：
<mrow>
<msub>
<mi>inh</mi>
<mrow>
<mi>d</mi>
<mi>y</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msubsup>
<mi>r</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mi>E</mi>
</msubsup>
<mo>&CenterDot;</mo>
<msub>
<mi>w</mi>
<mi>d</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>18</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein inhdy(x, y) is basic point being responded apart from weights to central nervous member of non-classical receptive field, wd(x, y) is non-warp
Distance weighting function of the allusion quotation receptive field to central nervous member；
Wherein, wdThe expression formula of (x, y) is：
<mrow>
<msub>
<mi>w</mi>
<mi>d</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>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>)</mo>
</mrow>
<mrow>
<mo>|</mo>
<mo>|</mo>
<mi>H</mi>
<mrow>
<mo>(</mo>
<mi>D</mi>
<mi>o</mi>
<mi>G</mi>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>)</mo>
<mo>|</mo>
<msub>
<mo>|</mo>
<mn>1</mn>
</msub>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>19</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>
<mi>k</mi>
<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>
<mi>k</mi>
<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 (DoG (x, y)) be take on the occasion of function, DoG (x, y) be the corresponding table of DoG templates
Up to formula；
Central nervous member is by the stimuli responsive expression formula of non-classical receptive field：
8. the profile testing method as claimed in claim 7 based on non-classical receptive field and linear non-linear modulation, its feature
It is：
Described step F is specific as follows：
For X cell：
Described central nervous member is by the expression formula of classical receptive field and the stimuli responsive of non-classical receptive field combined modulation：
Rx(x, y)=H (Ec (x, y)-α inhx(x,y)) (21)；
Wherein Rx(x, y) is that the corresponding central nervous member of X cell is stimulated by classical receptive field and non-classical receptive field combined modulation
Response, α is the corresponding rejection coefficient of X cell；
For Y cell：
Described central nervous member is by the expression formula of classical receptive field and the stimuli responsive of non-classical receptive field combined modulation：
Ry(x, y)=H (Ec (x, y)-β inhy(x,y)) (22)；
Wherein Ry(x, y) is that the corresponding central nervous member of Y cell is stimulated by classical receptive field and non-classical receptive field combined modulation
Response, β is the corresponding rejection coefficient of Y cell；
The described positive function that takes is
The profile value expression of the described corresponding pixel of central nervous member is as follows：
R (x, y)=Rx(x,y)+Ry(x,y) (23)；
The profile value of the corresponding pixel of neuron centered on wherein R (x, y).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710230469.8A CN107067407B (en) | 2017-04-11 | 2017-04-11 | Contour detection method based on non-classical receptive field and linear nonlinear modulation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710230469.8A CN107067407B (en) | 2017-04-11 | 2017-04-11 | Contour detection method based on non-classical receptive field and linear nonlinear modulation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107067407A true CN107067407A (en) | 2017-08-18 |
| CN107067407B CN107067407B (en) | 2020-01-31 |
Family
ID=59601812
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710230469.8A Active CN107067407B (en) | 2017-04-11 | 2017-04-11 | Contour detection method based on non-classical receptive field and linear nonlinear modulation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107067407B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210493A (en) * | 2019-04-30 | 2019-09-06 | 中南民族大学 | Profile testing method and system based on non-classical receptive field modulation neural network |
| 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 |
| CN111161291A (en) * | 2019-12-31 | 2020-05-15 | 广西科技大学 | Contour detection method based on target depth of field information |
| CN111179294A (en) * | 2019-12-30 | 2020-05-19 | 广西科技大学 | Bionic type contour detection method based on X, Y parallel visual channel response |
| CN113192092A (en) * | 2021-05-07 | 2021-07-30 | 广西科技大学 | Contour detection method for simulating fusion of characteristics of receptor field of XYW cells |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254304A (en) * | 2011-06-17 | 2011-11-23 | 电子科技大学 | Method for detecting contour of target object |
| US20140355861A1 (en) * | 2011-08-25 | 2014-12-04 | Cornell University | Retinal encoder for machine vision |
| CN104484667A (en) * | 2014-12-30 | 2015-04-01 | 华中科技大学 | Contour extraction method based on brightness characteristic and contour integrity |
| CN106033607A (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 |
| CN106033606A (en) * | 2015-07-24 | 2016-10-19 | 广西科技大学 | Target contour detection method of biomimetic smooth tracking eye movement information processing mechanism |
| CN106033610A (en) * | 2016-03-22 | 2016-10-19 | 广西科技大学 | Contour detection method based on non-classical receptive field space summation modulation |
| WO2016195496A2 (en) * | 2015-06-05 | 2016-12-08 | Universiteit Van Amsterdam | Deep receptive field networks |
| CN106228547A (en) * | 2016-07-15 | 2016-12-14 | 华中科技大学 | A kind of view-based access control model color theory and homogeneity suppression profile and border detection algorithm |
-
2017
- 2017-04-11 CN CN201710230469.8A patent/CN107067407B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254304A (en) * | 2011-06-17 | 2011-11-23 | 电子科技大学 | Method for detecting contour of target object |
| US20140355861A1 (en) * | 2011-08-25 | 2014-12-04 | Cornell University | Retinal encoder for machine vision |
| CN104484667A (en) * | 2014-12-30 | 2015-04-01 | 华中科技大学 | Contour extraction method based on brightness characteristic and contour integrity |
| WO2016195496A2 (en) * | 2015-06-05 | 2016-12-08 | Universiteit Van Amsterdam | Deep receptive field networks |
| CN106033607A (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 |
| CN106033606A (en) * | 2015-07-24 | 2016-10-19 | 广西科技大学 | Target contour detection method of biomimetic smooth tracking eye movement information processing mechanism |
| CN106033610A (en) * | 2016-03-22 | 2016-10-19 | 广西科技大学 | Contour detection method based on non-classical receptive field space summation modulation |
| CN106228547A (en) * | 2016-07-15 | 2016-12-14 | 华中科技大学 | A kind of view-based access control model color theory and homogeneity suppression profile and border detection algorithm |
Non-Patent Citations (4)
| Title |
|---|
| BASABI BHAUMIK 等: "A Cooperation and Competition Based Simple Cell Receptive Field Model and Study of Feed-Forward Linear and Nonlinear Contributions to Orientation Selectivity", 《COMPUTATIONAL NEUROSCIENCE》 * |
| CHUAN LIN 等: "Improved contour detection model with spatial summation properties based on nonclassical receptive field", 《J. OF ELECTRONIC IMAGING》 * |
| 李康群 等: "基于视通路多感受野朝向性关联的轮廓检测方法", 《中国生物医学工程学报》 * |
| 谢昭: "一种仿生物视觉感知的视频轮廓检测方法", 《自动化学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210493A (en) * | 2019-04-30 | 2019-09-06 | 中南民族大学 | Profile testing method and system based on non-classical receptive field modulation neural network |
| CN110210493B (en) * | 2019-04-30 | 2021-03-19 | 中南民族大学 | Contour detection method and system based on non-classical receptive field modulation neural network |
| 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 |
| CN111179294A (en) * | 2019-12-30 | 2020-05-19 | 广西科技大学 | Bionic type contour detection method based on X, Y parallel visual channel response |
| CN111161291A (en) * | 2019-12-31 | 2020-05-15 | 广西科技大学 | Contour detection method based on target depth of field information |
| CN113192092A (en) * | 2021-05-07 | 2021-07-30 | 广西科技大学 | Contour detection method for simulating fusion of characteristics of receptor field of XYW cells |
| CN113192092B (en) * | 2021-05-07 | 2023-04-07 | 广西科技大学 | Contour detection method for simulating fusion of properties of receptor field of XYW cell |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107067407B (en) | 2020-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107067407A (en) | Profile testing method based on non-classical receptive field and linear non-linear modulation | |
| CN106033610B (en) | Profile testing method based on the modulation of non-classical receptive field spatial summation | |
| Xia et al. | Spectral–spatial classification of hyperspectral images using ICA and edge-preserving filter via an ensemble strategy | |
| CN107767387A (en) | Profile testing method based on the global modulation of changeable reception field yardstick | |
| CN103177458B (en) | A kind of visible remote sensing image region of interest area detecting method based on frequency-domain analysis | |
| CN106845421A (en) | Face characteristic recognition methods and system based on multi-region feature and metric learning | |
| CN106446942A (en) | Crop disease identification method based on incremental learning | |
| CN106991368A (en) | A kind of finger vein checking personal identification method based on depth convolutional neural networks | |
| CN105744256A (en) | Three-dimensional image quality objective evaluation method based on graph-based visual saliency | |
| CN106033608B (en) | The objective contour detection method of bionical object smooth pursuit eye movement information processing mechanism | |
| CN103839269A (en) | Image segmentation method based on quaternion and fuzzy C-means clustering | |
| CN108053398A (en) | A kind of melanoma automatic testing method of semi-supervised feature learning | |
| CN106780582A (en) | Based on the image significance detection method that textural characteristics and color characteristic are merged | |
| CN107451594A (en) | A kind of various visual angles Approach for Gait Classification based on multiple regression | |
| CN100456319C (en) | High spectrum image repeated selection weighing classification method | |
| CN106127740A (en) | A kind of profile testing method based on the association of visual pathway many orientation of sensory field | |
| CN106971161A (en) | Face In vivo detection system based on color and singular value features | |
| CN106033609B (en) | The objective contour detection method of bionical object jump eye movement information processing mechanism | |
| CN101533475A (en) | Method for extracting feature of shape-adaptive neighborhood based remote sensing image | |
| CN104463821A (en) | Method for fusing infrared image and visible light image | |
| CN116385819A (en) | Water quality evaluation method, device and equipment based on neural network model | |
| CN107742302A (en) | Profile testing method based on the multiple dimensioned profile fusion of primary visual cortex | |
| CN107766866A (en) | Set direction profile testing method based on receptive field subregion | |
| CN101859381A (en) | Gabor filter, image recognition apparatus and method, program and recording medium | |
| CN106296749A (en) | RGB D image eigen decomposition method based on L1 norm constraint |
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 |