CN108053415B - Bionic contour detection method based on improved non-classical receptive field - Google Patents

Bionic contour detection method based on improved non-classical receptive field Download PDF

Info

Publication number
CN108053415B
CN108053415B CN201711333790.5A CN201711333790A CN108053415B CN 108053415 B CN108053415 B CN 108053415B CN 201711333790 A CN201711333790 A CN 201711333790A CN 108053415 B CN108053415 B CN 108053415B
Authority
CN
China
Prior art keywords
pixel point
receptive field
response
classical receptive
function
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201711333790.5A
Other languages
Chinese (zh)
Other versions
CN108053415A (en
Inventor
林川
李福章
张晴
曹以隽
韦艳霞
潘勇才
刘青正
张玉薇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangxi University of Science and Technology
Original Assignee
Guangxi University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangxi University of Science and Technology filed Critical Guangxi University of Science and Technology
Priority to CN201711333790.5A priority Critical patent/CN108053415B/en
Publication of CN108053415A publication Critical patent/CN108053415A/en
Application granted granted Critical
Publication of CN108053415B publication Critical patent/CN108053415B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention aims to provide a bionic contour detection method based on an improved non-classical receptive field, which comprises the following steps: A. inputting a to-be-detected image subjected to gray level processing; B. presetting a Gabor filter function containing a plurality of direction parameters, respectively carrying out classical receptive field stimulation response on each pixel point in an image to be detected, and taking the corresponding direction as the optimal direction of the pixel point; C. constructing an inhibition kernel function by using a log function, and constructing a distance weight function by using the inhibition kernel function; convolving the classical receptive field stimulus response of each pixel point with the distance weight function of the pixel point to obtain the inhibition response of the pixel point; D. and subtracting the product of the suppression response of the pixel point and the preset suppression strength from the classical receptive field stimulation response of each pixel point, and calculating to obtain the final contour value of the pixel point. The detection method overcomes the defects of the prior art, and has the characteristics of meeting the spatial characteristics of visual receptive field and having better detection effect.

Description

Bionic contour detection method based on improved non-classical receptive field
Technical Field
The invention relates to the field of image processing, in particular to a bionic contour detection method based on an improved non-classical receptive field.
Background
For non-classical receptive field regions, there are many models of different inhibition situations. On the basis of the DoG suppression model, the traditional butterfly suppression model completes the suppression of background texture information by artificially limiting the areas of a side area and an end area; the butterfly-shaped inhibition nucleus is divided into regions manually, firstly, a non-classical receptive field needs to be divided into a flank region and a top region through a +/-45-degree alignment line, and therefore a butterfly-shaped inhibition model of the non-automatic defined region is obtained. In the butterfly-type suppression model, the side region and the end region have different working rules, namely the calculation of the suppression strength of the side region is completed based on an accurate scale feature, and the calculation of the suppression strength of the end region is adaptively changed along with local features on different spatial scales. In local texture regions where significant contouring is likely to exist, the end regions will impose a weaker suppression of the classical field, and conversely, in regions filled with random texture elements, a stronger suppression. The traditional butterfly-type suppression model cannot realize the automatic division of the suppression region and the reasonable rotation of the suppression model by utilizing the optimal direction of the pixel points, so the time cost of the whole contour detection process is greatly increased.
Disclosure of Invention
The invention aims to provide a bionic contour detection method based on an improved non-classical receptive field, which overcomes the defects of the prior art and has the characteristics of good detection effect and high detection efficiency.
The technical scheme of the invention is as follows:
a bionic contour detection method based on improved non-classical receptive fields comprises the following steps:
A. inputting a to-be-detected image subjected to gray level processing;
B. presetting a Gabor filter function containing a plurality of direction parameters, and performing Gabor energy calculation on each pixel point in an image to be detected by using the Gabor filter function respectively to obtain Gabor energy values of each pixel point in each direction; for each pixel point, selecting the maximum value of the Gabor energy values in each direction as the classical receptive field stimulation response of the pixel point, and taking the direction corresponding to the maximum value as the optimal direction of the pixel point;
C. constructing an inhibition kernel function, and constructing a distance weight function through the inhibition kernel function; convolving the classical receptive field stimulus response of each pixel point with the distance weight function of the pixel point to obtain the inhibition response of the pixel point;
the inhibiting kernel function log (x, y; epsilon, sigma)w) Comprises the following steps:
Figure BDA0001507090590000011
wherein
Figure BDA0001507090590000021
Figure BDA0001507090590000022
For the optimal direction of pixel (x, y), ε is 0.1, σwTo suppress the nuclear scale;
D. and subtracting the product of the suppression response of the pixel point and the preset suppression strength from the classical receptive field stimulation response of each pixel point to obtain the contour response of the pixel point, and performing non-maximum suppression and double-threshold processing on the contour response to obtain the final contour value of each pixel point.
Preferably, the step B is as follows:
the expression of the Gabor filter function is as follows:
Figure BDA0001507090590000023
wherein
Figure BDA0001507090590000024
Gamma is a constant representing the ratio of the long axis to the short axis of the elliptical receptive field, the parameter lambda is the wavelength, sigma is the scale, 1/lambda is the spatial frequency of the cosine function,
Figure BDA0001507090590000025
is a phase angle parameter, theta is a direction parameter of Gabor filtering;
the Gabor energy value was calculated as follows:
Figure BDA0001507090590000026
wherein
Figure BDA0001507090590000027
Figure BDA0001507090590000028
Wherein theta isiFor a certain directional parameter of Gabor filtering, NθFor directional parameters of Gabor filteringThe number of the cells; i (x, y) is an image to be detected and is a convolution operator;
the expression of the classical receptive field stimulus response Ec (x, y) is as follows:
Figure BDA0001507090590000029
Figure BDA00015070905900000210
preferably, the step C specifically includes:
the distance weight function wσ(x,y；ε,σw) Comprises the following steps:
Figure BDA00015070905900000211
wherein,
Figure BDA0001507090590000031
wherein | · | purple1Is (L)1) Norm, h (x) max (0, x);
the suppression response Inh (x, y) of each pixel point is as follows:
Inh(x,y)＝Ec(x,y)*wσ(x,y；ε,σw) (8)。
preferably, the step D specifically includes:
the profile response R (x, y) is:
R(x,y)＝H(Ec(x,y)-αInh(x,y)) (9)；
where h (x) max (0, x), α is the inhibition intensity.
According to the method, the improved suppression model is adopted, the non-classical receptive field region does not need to be manually divided, the rotation of the suppression region is carried out in the optimal direction through the log function suppression core with the angle parameter, the workload of manual division is saved, and the overall operation efficiency is improved; the improved inhibition model enables the inhibition of each pixel point to be carried out in combination with the optimal direction of the pixel point, so that the optimal inhibition effect is achieved, and meanwhile, the visual characteristics of the receptive field are better met.
Drawings
FIG. 1 is a graph showing a comparison of the detection results of the contour detection method of example 1 and the method of reference 1
Detailed Description
The present invention will be described in detail with reference to the accompanying drawings and examples.
Example 1
A bionic contour detection method based on improved non-classical receptive fields comprises the following steps:
A. inputting a to-be-detected image subjected to gray level processing;
B. presetting a Gabor filter function containing a plurality of direction parameters, and performing Gabor energy calculation on each pixel point in an image to be detected by using the Gabor filter function respectively to obtain Gabor energy values of each pixel point in each direction; for each pixel point, selecting the maximum value of the Gabor energy values in each direction as the classical receptive field stimulation response of the pixel point, and taking the direction corresponding to the maximum value as the optimal direction of the pixel point;
the step B is specifically as follows:
the expression of the Gabor filter function is as follows:
Figure BDA0001507090590000032
wherein
Figure BDA0001507090590000041
Gamma is a constant representing the ratio of the long axis to the short axis of the elliptical receptive field, the parameter lambda is the wavelength, sigma is the scale, 1/lambda is the spatial frequency of the cosine function,
Figure BDA0001507090590000042
is a phase angle parameter, theta is a direction parameter of Gabor filtering;
the Gabor energy value was calculated as follows:
Figure BDA0001507090590000043
wherein
Figure BDA0001507090590000044
Figure BDA00015070905900000412
Wherein theta isiFor a certain directional parameter of Gabor filtering, NθThe number of the direction parameters of Gabor filtering is obtained; i (x, y) is an image to be detected and is a convolution operator;
the expression of the classical receptive field stimulus response Ec (x, y) is as follows:
Figure BDA0001507090590000045
Figure BDA0001507090590000046
wherein
Figure BDA0001507090590000047
The optimal direction of the pixel point (x, y);
C. constructing an inhibition kernel function, and constructing a distance weight function through the inhibition kernel function; convolving the classical receptive field stimulus response of each pixel point with the distance weight function of the pixel point to obtain the inhibition response of the pixel point;
the inhibiting kernel function log (x, y; epsilon, sigma)w) Comprises the following steps:
Figure BDA0001507090590000048
wherein
Figure BDA0001507090590000049
ε＝0.1，σwTo suppress the nuclear scale;
the distance weight function wσ(x,y；ε,σw) Comprises the following steps:
Figure BDA00015070905900000410
wherein,
Figure BDA00015070905900000411
wherein, epsilon is 0.1, sigmaw＝σ，||·||1Is (L)1) Norm, h (x) max (0, x);
the suppression response Inh (x, y) of each pixel point is as follows:
Inh(x,y)＝Ec(x,y)*wσ(x,y；ε,σw) (10)；
D. subtracting the product of the suppression response of each pixel point and the preset suppression strength from the classical receptive field stimulation response of each pixel point to obtain the contour response of the pixel point, and performing non-maximum suppression and double-threshold processing on the contour response to obtain the final contour value of each pixel point;
the step D is specifically as follows:
the profile response R (x, y) is:
R(x,y)＝H(Ec(x,y)-αInh(x,y)) (11)；
where h (x) max (0, x), α is the inhibition intensity;
the non-maximum suppression and binarization processing according to the present embodiment employs the method described in document 1, in which two threshold values t are includedh,tlIs set to tl＝0.5thCalculated from a threshold quantile p;
document 1: GrigoresecuC, PetkovN, WestenbergM. Contourdiscionbasedonnotronfibrous receiptionJ. IEEETransactionson Imageprocessing,2003,12(7): 729-;
the effectiveness of the contour detection method of the present embodiment is compared with the contour detection isotropy model provided in document 1, where the performance evaluation index P adopts the following criteria given in document 1:
Figure BDA0001507090590000051
in the formula nTP、nFP、nFNRespectively representing the number of detected correct contours, error contours and missing contours, and the evaluation index P is set to be [0,1 ]]The closer to 1, the better the contour detection, and in addition, the tolerance is defined as: all detected within 5 x 5 neighbourhoods are counted as correct detections;
selecting 4 pairs of classical images in fig. 1 for effectiveness comparison, and respectively adopting an isotropic model in literature 1 and an embodiment 1 method to perform contour detection on the 4 images, wherein the parameter set selected by the embodiment 1 method is shown in table 1, and the optimal results obtained from the parameter set are selected for comparison;
table 1 example 1 parameter set table
Figure BDA0001507090590000052
Figure BDA0001507090590000061
The isotropic model in document 1 uses α ═ 1.0,1.2 ═ 1.4,1.6,1.8,2.0,2.2,2.4,2.6,2.8 ═ 0.1,0.2,0.3,0.4,0.5 };
fig. 1 shows an original image, an actual contour map, and an optimal contour detected by the method of document 1, which are 4 pairs of classical images including basket, elephant _2, goat _3, and hyena, respectively, and the optimal contour detected by the method of embodiment 1; as shown in table 2, the comparison between the optimal P value detected by the method of document 1 and the optimal P value detected by the method of example 1 is shown for the 4 images;
TABLE 2 optimal P-value comparison
Figure BDA0001507090590000062
From the above results, it can be seen that the method of example 1 is superior to the isotropic model in document 1 both in the effect of contour extraction and in the performance index parameter.

Claims (4)

1. A bionic contour detection method based on an improved non-classical receptive field is characterized by comprising the following steps:
A. inputting a to-be-detected image subjected to gray level processing;
B. presetting a Gabor filter function containing a plurality of direction parameters, and performing Gabor energy calculation on each pixel point in an image to be detected by using the Gabor filter function respectively to obtain Gabor energy values of each pixel point in each direction; for each pixel point, selecting the maximum value of the Gabor energy values in each direction as the classical receptive field stimulation response of the pixel point, and taking the direction corresponding to the maximum value as the optimal direction of the pixel point;
C. constructing an inhibition kernel function, and constructing a distance weight function through the inhibition kernel function; convolving the classical receptive field stimulus response of each pixel point with the distance weight function of the pixel point to obtain the inhibition response of the pixel point;
the inhibiting kernel function log (x, y; epsilon, sigma)w) Comprises the following steps:
Figure DEST_PATH_FDA0001562917900000011
wherein
Figure DEST_PATH_FDA0001562917900000012
Figure DEST_PATH_FDA0001562917900000013
For the optimal direction of pixel (x, y), ε is 0.1, σwTo suppress the nuclear scale;
D. and subtracting the product of the suppression response of the pixel point and the preset suppression strength from the classical receptive field stimulation response of each pixel point to obtain the contour response of the pixel point, and performing non-maximum suppression and double-threshold processing on the contour response to obtain the final contour value of each pixel point.
2. The improved non-classical receptive field-based biomimetic contour detection method according to claim 1, characterized in that:
the step B is specifically as follows:
the expression of the Gabor filter function is as follows:
Figure DEST_PATH_FDA0001562917900000014
wherein
Figure DEST_PATH_FDA0001562917900000015
Gamma is a constant representing the ratio of the long axis to the short axis of the elliptical receptive field, the parameter lambda is the wavelength, sigma is the scale, 1/lambda is the spatial frequency of the cosine function,
Figure DEST_PATH_FDA0001562917900000016
is a phase angle parameter, theta is a direction parameter of Gabor filtering;
the Gabor energy value was calculated as follows:
Figure DEST_PATH_FDA0001562917900000017
wherein
Figure DEST_PATH_FDA0001562917900000018
Figure DEST_PATH_FDA0001562917900000019
Wherein theta isiFor a certain directional parameter of Gabor filtering, NθThe number of the direction parameters of Gabor filtering is obtained; i (x, y) is an image to be detected and is a convolution operator;
the expression of the classical receptive field stimulus response Ec (x, y) is as follows:
Figure DEST_PATH_FDA0001562917900000021
Figure DEST_PATH_FDA0001562917900000022
3. the improved non-classical receptive field-based biomimetic contour detection method according to claim 2, characterized in that:
the step C is specifically as follows:
the distance weight function wσ(x,y；ε,σw) Comprises the following steps:
Figure DEST_PATH_FDA0001562917900000023
wherein,
Figure DEST_PATH_FDA0001562917900000024
wherein | · | purple1Is (L)1) Norm, h (x) max (0, x);
the suppression response Inh (x, y) of each pixel point is as follows:
Inh(x,y)＝Ec(x,y)*wσ(x,y；ε,σw) (8)。
4. the improved non-classical receptive field-based biomimetic contour detection method according to claim 3, characterized in that:
the step D is specifically as follows:
the profile response R (x, y) is:
R(x,y)＝H(Ec(x,y)-αInh(x,y)) (9)；
where h (x) max (0, x), α is the inhibition intensity.
CN201711333790.5A 2017-12-14 2017-12-14 Bionic contour detection method based on improved non-classical receptive field Active CN108053415B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711333790.5A CN108053415B (en) 2017-12-14 2017-12-14 Bionic contour detection method based on improved non-classical receptive field

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711333790.5A CN108053415B (en) 2017-12-14 2017-12-14 Bionic contour detection method based on improved non-classical receptive field

Publications (2)

Publication Number Publication Date
CN108053415A CN108053415A (en) 2018-05-18
CN108053415B true CN108053415B (en) 2020-05-22

Family

ID=62132689

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711333790.5A Active CN108053415B (en) 2017-12-14 2017-12-14 Bionic contour detection method based on improved non-classical receptive field

Country Status (1)

Country Link
CN (1) CN108053415B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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
CN109978898B (en) * 2019-02-01 2023-07-18 广西科技大学 Contour detection method based on vector field energy calculation
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
CN111968141B (en) * 2020-06-30 2023-06-16 广西科技大学 Non-classical receptive field contour detection method based on multi-scale superposition

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5740274A (en) * 1991-09-12 1998-04-14 Fuji Photo Film Co., Ltd. Method for recognizing object images and learning method for neural networks
US7433540B1 (en) * 2002-10-25 2008-10-07 Adobe Systems Incorporated Decomposing natural image sequences
CN101286237B (en) * 2008-05-22 2010-04-14 重庆大学 Movement target detection method based on visual sense bionics
CN102254304B (en) * 2011-06-17 2012-08-29 电子科技大学 Method for detecting contour of target object
CN102510368B (en) * 2012-01-04 2014-04-16 西安电子科技大学 Wireless orthogonal frequency division multiplexing (OFDM) signal peak-to-average ratio inhibition method based on amplitude distribution variation
CN102750731B (en) * 2012-07-05 2016-03-23 北京大学 Based on the remarkable computing method of stereoscopic vision of the simple eye receptive field in left and right and binocular fusion
CN104318550A (en) * 2014-09-27 2015-01-28 励盼攀 Eight-channel multi-spectral imaging data processing method
CN104484667B (en) * 2014-12-30 2018-02-23 华中科技大学 A kind of contour extraction method based on brightness and integrality of outline
CN106033608B (en) * 2015-07-24 2018-10-19 广西科技大学 The objective contour detection method of bionical object smooth pursuit eye movement information processing mechanism
CN105261015B (en) * 2015-09-29 2018-01-12 北京工业大学 Eye fundus image blood vessel automatic division method based on Gabor filter
CN105654496B (en) * 2016-01-08 2018-06-29 华北理工大学 The bionical adaptive fuzzy edge detection method of view-based access control model characteristic
CN106033610B (en) * 2016-03-22 2018-10-26 广西科技大学 Profile testing method based on the modulation of non-classical receptive field spatial summation

Also Published As

Publication number Publication date
CN108053415A (en) 2018-05-18

Similar Documents

Publication Publication Date Title
CN108053415B (en) Bionic contour detection method based on improved non-classical receptive field
CN107767387B (en) Contour detection method based on variable receptive field scale global modulation
CN104200442B (en) Non-local mean MRI image denoising method based on improved canny rim detections
CN103116875B (en) Self-adaptation bilateral filtering image de-noising method
CN106033610B (en) Profile testing method based on the modulation of non-classical receptive field spatial summation
CN106204447A (en) The super resolution ratio reconstruction method with convolutional neural networks is divided based on total variance
CN103955922A (en) Method for detecting flaws of printed fabric based on Gabor filter
CN107742289A (en) One kind is based on machine vision revolving body workpieces detection method
CN107909593B (en) Non-directional selective contour detection method based on receptive field region
CN107220988A (en) Based on the parts image edge extraction method for improving canny operators
CN105205828B (en) Knitted fabric flaw detection method based on Optimal Gabor Filters
CN103914827A (en) Method for visual inspection of shortages of automobile sealing strip profile
CN103839234A (en) Double-geometry nonlocal average image denoising method based on controlled nuclear
CN107845086A (en) A kind of detection method, system and the device of leather surface conspicuousness defect
CN102306378B (en) Image enhancement method
CN107766866B (en) Direction selective contour detection method based on receptive field region
CN105225243B (en) One kind can antimierophonic method for detecting image edge
CN107369144A (en) Based on the multiple dimensioned Retinex images defogging method for improving homomorphic filtering
CN102663706A (en) Adaptive weighted mean value filtering method based on diamond template
CN114820401B (en) Method for enhancing marine backlight infrared image by combining histogram transformation and edge information
CN105427262A (en) Image de-noising method based on bidirectional enhanced diffusion filtering
CN107742302B (en) Contour detection method based on primary visual cortex multi-scale contour fusion
CN109978898B (en) Contour detection method based on vector field energy calculation
CN103903227B (en) Method and device for noise reduction of image
CN108090492B (en) Contour detection method based on scale clue suppression

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
EE01 Entry into force of recordation of patent licensing contract

Application publication date: 20180518

Assignee: GUANGXI YINGTENG EDUCATION TECHNOLOGY Co.,Ltd.

Assignor: GUANGXI University OF SCIENCE AND TECHNOLOGY

Contract record no.: X2023980053979

Denomination of invention: A biomimetic contour detection method based on improved non classical receptive field

Granted publication date: 20200522

License type: Common License

Record date: 20231226