CN117593323A - Image segmentation method, system, medium and device based on non-local features - Google Patents
Image segmentation method, system, medium and device based on non-local features Download PDFInfo
- Publication number
- CN117593323A CN117593323A CN202410079491.7A CN202410079491A CN117593323A CN 117593323 A CN117593323 A CN 117593323A CN 202410079491 A CN202410079491 A CN 202410079491A CN 117593323 A CN117593323 A CN 117593323A
- Authority
- CN
- China
- Prior art keywords
- image
- local
- level set
- segmented
- contour line
- 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
- 238000003709 image segmentation Methods 0.000 title claims abstract description 64
- 238000000034 method Methods 0.000 title claims abstract description 55
- 230000011218 segmentation Effects 0.000 claims abstract description 39
- 238000010606 normalization Methods 0.000 claims abstract description 27
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 10
- 238000010586 diagram Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 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
- G06T7/12—Edge-based segmentation
-
- 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
- G06T2207/20112—Image segmentation details
- G06T2207/20161—Level set
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to an image segmentation method, system, medium and equipment based on non-local features, wherein the method comprises the following steps: acquiring an image to be segmented, and setting an initial contour line in the image to be segmented by using a level set method; establishing a non-local feature driving item by combining the non-local feature image and a scalable normalization method; iteratively updating the initial level set function by using the non-local characteristic driving term to obtain a final contour line; and segmenting the image to be segmented by using the final contour line to obtain an image segmentation result. According to the method and the device, the non-local characteristic driving term is used for carrying out iterative updating on the initial level set function, so that a segmentation result is obtained, the image segmentation efficiency and the segmentation precision are improved, and the method and the device are suitable for segmentation of the weak edge and the noise image.
Description
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to a method, a system, a medium, and an apparatus for image segmentation based on non-local features.
Background
Image segmentation is an important research direction in the fields of computer vision and image processing. Image segmentation is the process of dividing an image into several disjoint connected sub-regions and extracting a region of interest (ROI).
The active contour model is a widely used class of algorithms in image segmentation algorithms. The basic idea of the active contour model is to implicitly represent the contour line as a zero level set of a level set function, and the initial contour line evolves to the target boundary under the iterative update of the level set function to obtain the image segmentation result. The algorithm can be used for constructing an energy function by combining various priori knowledge, sub-pixel precision can be achieved for the boundary of the target to be segmented, and the segmentation result is a smooth closed contour.
In the active contour model, the level set function is iteratively updated to evolve towards the target boundary under the action force of the driving term, and the ideal driving term should only reflect the target boundary, however, due to the diversity of image types and image contrast, the driving term obtained by calculation is difficult to accurately extract image features and reflect accurate target boundary features, so that the accurate segmentation result is not easy to obtain. In addition, in the active contour model of the existing mainstream, the calculation of the driving term needs to be updated at each level set iteration, resulting in low segmentation efficiency.
Disclosure of Invention
Therefore, the invention aims to solve the technical problems of inaccurate image feature extraction, inaccurate target boundary feature reflection and low segmentation efficiency of the driving item in the prior art. In order to solve the technical problems, the invention provides an image segmentation method, an image segmentation system, an image segmentation medium and an image segmentation device based on non-local features, which can improve the image segmentation efficiency and the segmentation precision and are suitable for segmenting weak edges and noise images.
In order to achieve the above purpose, the main technical scheme adopted by the invention comprises the following steps:
in a first aspect, the present invention provides a method for image segmentation based on non-local features, the method comprising: acquiring an image to be segmented, and setting an initial contour line in the image to be segmented by using a level set method; establishing a non-local feature driving item by combining the non-local feature image and a scalable normalization method; iteratively updating the initial level set function by using the non-local characteristic driving term to obtain a final contour line; and segmenting the image to be segmented by using the final contour line to obtain an image segmentation result.
In one embodiment of the present invention, an initial contour line is set in the image to be segmented using a level set method, which is specifically as follows: representing the initial contour line by using the initial level set function, wherein when the vector pixel point of the image to be segmented is outside the initial contour line, the value of the initial level set function is 1; when the vector pixel point of the image to be segmented is in the initial contour line or on the initial contour line, the value of the initial level set function is-1.
In one embodiment of the present invention, the method for creating the non-local feature driver is expressed as:
;
wherein,for vector pixel points in the image to be segmented, < > in->For the non-local feature driven term,for determining the direction sign of the curve evolution direction +.>To adjust the constant of the amplitude of the non-local feature driver term,for a scalable normalization function, +.>Is a telescopic factor->Gray values of the image that are non-local feature differences;
the direction symbol of the evolution direction of the judgment curve is expressed as follows:
;
wherein,representing the initial level set function, +.>Representing an average calculation function, +.>Average gray value of vector pixel points representing the value of the initial level set function less than 0,/v>Representing the average gray value of the vector pixel points with the value of the initial level set function being more than or equal to 0;
the scalable normalization function is expressed as:
;
wherein,a calculation variable representing said scalable normalization function, -a calculation variable representing said scalable normalization function>Is a telescopic factor->Is->The abscissa of the positive turning point,/>to distinguish between the threshold of the edge and non-edge regions of the drive term.
In one embodiment of the present invention, the gray value of the image of the non-local feature difference is expressed as:
;
wherein,for images processed by the non-local mean algorithm, < >>Non-local weighted fitting images for the image to be segmented for fitting foreground and background,/for the image to be segmented>As a logarithmic function>The standard deviation of the gray scale of the image to be segmented is obtained;
the non-local mean algorithm processed image is expressed as:
;
wherein,to->Is the center and the side center distance is +.>Square window of>,/>For smooth parameters +.>To lower round symbol->For window->Vector pixel points of (2); />For the gray value of the image to be segmented, < >>Is->And->Non-local weights in between.
In one embodiment of the invention, the non-local weighted fit image comprises:
a non-local weighted fit image of the high gray sub-region, expressed as:
;
a non-local weighted fit image of the low gray sub-region, expressed as:
;
wherein,for sub-areas where the image gray values are higher than the average gray value,for images having gray values lower than the average gray valueSubregion (S)>For window->Average gray values of all vector pixels.
In one embodiment of the invention, the iterative update is performed using the following formula:
;
wherein,for the number of iterations->For normalization function->Is a constant, +.>Representing the initial level set iteration +.>Post-level set function ++>Representing the initial level set iteration +.>Post-level set function ++>To->Is an average filter of a convolution kernel template, +.>Is +.>,/>Is a constant, +.>For convolution operation symbol>For the time step +.>To fit the dirac function, +.>Is a constant.
In one embodiment of the present invention, in the step of iteratively updating the initial level set function using the non-local feature driver term, when a termination condition is satisfiedWhen the level set function stops iterative updating, and the last updated level set function is used as the final contour line, wherein +_>Is a termination condition coefficient.
In a second aspect, the present invention provides an image segmentation system based on non-local features, comprising:
the device comprises an initial contour line setting module, a contour line segmentation module and a contour line segmentation module, wherein the initial contour line setting module is used for acquiring an image to be segmented and setting an initial contour line in the image to be segmented by using a level set method;
the non-local feature driving item establishing module is used for establishing a non-local feature driving item by combining the non-local feature image and a scalable normalization method;
the final contour line generation module is used for carrying out iterative updating on the initial level set function by using the non-local characteristic driving term to obtain a final contour line;
and the image segmentation module is used for segmenting the image to be segmented by using the final contour line to obtain an image segmentation result.
In a third aspect, the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed, implements the non-local feature based image segmentation method of any of the first aspects above.
In a fourth aspect, the present invention provides a storage device comprising a storage medium and a processor, the storage medium storing a computer program which, when executed by the processor, implements the non-local feature based image segmentation method according to any one of the first aspects.
Advantageous effects
The beneficial effects of the invention are as follows: according to the image segmentation method, system, medium and equipment based on the non-local features, the non-local feature driving item is established by combining the non-local feature image and the scalable normalization method, the improved non-local feature driving item is used for carrying out iterative updating on the initial level set function, and the non-local feature driving item does not participate in the level set iterative updating process but is calculated before the level set iteration, so that the influence of the initial contour line on the segmentation precision is small, the segmentation robustness is enhanced, the calculation complexity in the image segmentation is reduced, and the segmentation efficiency is improved; meanwhile, the similarity of structural features of non-local areas of the image is considered in the non-local feature image, so that the non-local feature driving item can accurately extract image features and reflect target boundary features, and the image processed by the non-local mean algorithm realizes noise reduction processing, so that the influence of noise on segmentation can be reduced; in addition, the scalable normalization method can enhance the capability of the driving term to distinguish the target from the background, enhance the robustness of segmentation, and further improve the accuracy of image segmentation.
Drawings
FIG. 1 is a schematic overall flow chart of a non-local feature-based image segmentation method according to a preferred embodiment of the present invention;
FIG. 2 is a schematic diagram of a result of segmenting a real image by the non-local feature-based image segmentation method shown in FIG. 1;
FIG. 3 is a schematic diagram showing the image segmentation method based on non-local features shown in FIG. 1, using different models to segment the image;
fig. 4 is a block diagram of an image segmentation system based on non-local features according to an embodiment of the present invention.
[ reference numerals description ]
600: an image segmentation system based on non-local features;
601: an initial contour setting module;
602: a non-local feature driver creation module;
603: a final contour line generation module;
604: and an image segmentation module.
Detailed Description
The invention will be better explained by the following detailed description of the embodiments with reference to the drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
In a first aspect, referring to fig. 1, a preferred embodiment of the present invention provides a non-local feature-based image segmentation method, including:
s1, acquiring an image to be segmented, and setting an initial contour line in the image to be segmented by using a level set method;
specifically, an initial level set function is used to represent an initial contour of an image to be segmented, the initial level set function being expressed as:
(1);
wherein,to be divided intoVector pixel point in image, +.>As a function of the initial level set.
When the vector pixel point of the image to be segmented is outside the initial contour line, the value of the initial level set function is 1; when the vector pixel point of the image to be segmented is inside or on the initial contour line, the value of the initial level set function is-1.
S2, establishing a non-local feature driving item by combining the non-local feature image and a scalable normalization method;
specifically, S2 includes:
setting a direction sign of a judging curve evolution direction, wherein the direction sign of the judging curve evolution direction is expressed as follows:
(2);
wherein,representing an average calculation function, +.>Average gray value of vector pixel points representing the value of the initial level set function less than 0,/>Average gray values of vector pixel points having a value equal to or greater than 0 represent the initial level set function.
Setting a constant for adjusting the amplitude of the non-local feature driving termIn this embodiment, < > a->The value range of (2) is [1,3 ]]。
Calculating standard deviation of gray scale of image to be segmentedStandard deviation of gray scale of the image to be segmented +.>Expressed as:
(3);
wherein,for the number of lines of the image to be segmented, < > for>For the number of columns of the image to be segmented +.>For the sum formula>For the region of the image to be segmented +.>For the gray value of the image to be segmented, +.>The gray average value of the image to be segmented.
Calculating the average gradient of the image to be segmentedThe average gradient of the image to be segmented +.>Expressed as:
(4);
wherein,is the gradient magnitude of the image to be segmented.
Setting smoothing parametersThe smoothing parameter->Expressed as:
(5);
wherein,to lower round symbol->As a logarithmic function; />Is a preset constant, in this embodiment +.>。
Is provided withSquare window as center->The edge-center distance of (2) is:
(6);
calculating a non-local mean algorithm processed image, the non-local mean algorithm processed image represented as:
(7);
wherein,for window->Vector pixel of>The gray value of the image to be segmented;is->And->Non-local weights in between, wherein +.>As an exponential function +.>、/>Respectively by、/>Is the center and the side center distance is +.>Is a square block of (a); />Is->And->Square Euclidean distance between->For traversing->And->Position vector of>Is a two-dimensional integer set->Is an infinite norm of the vector.
In this embodiment, the center-to-edge distance of the square block。
The method for calculating the non-local weighted fitting image of the image to be segmented, which is used for fitting the foreground and the background, specifically comprises the following steps:
defining the foreground and the background of the image to be segmented as a high gray level sub-region and a low gray level sub-region, and respectively representing non-local weighted fitting images of the high gray level sub-region and the low gray level sub-region as follows:
(8);
(9);
wherein,for sub-areas where the image gray values are higher than the average gray value,for sub-areas where the grey value of the image is lower than the average grey value,/->For window->Average gray values of all vector pixels.
Calculating a gray value of an image of a non-local feature difference, the gray value of the image of the non-local feature difference being expressed as:
(10);
wherein,for images processed by the non-local mean algorithm, < >>The image is fitted for non-local weighting of the image to be segmented for fitting the foreground and the background.
Setting a threshold value for distinguishing the edge and non-edge regions of the driving term, and calculating by the following formula:
(11);
wherein,is a variable of the summation formula; />Is an auxiliary function->First->Value of stage>To round up the whole symbol +.>Taking absolute value symbols; />To be +.>The number of points in (2); />Is an adjustable parameter, in this embodiment,/->The value range of (5) is [0.85,0.95 ]]。
Setting a scalable normalization functionThe scalable normalization function->Expressed as:
(12);
wherein,a computational variable representing a scalable normalization function, +.>Is a telescopic factor->Is->The abscissa of the positive turning point, in this embodiment +.>。
Establishing non-local feature driver entriesThe non-local feature driver->Expressed as:
(13);
wherein,is a vector pixel point in the image to be segmented.
The non-local feature image in the application considers the similarity of structural features of non-local areas of the image, and is a collective term for images processed by a non-local mean algorithm and non-local weighted fit images used for fitting foreground and background. The image processed by the non-local mean algorithm realizes noise reduction processing, and can reduce the influence of noise on segmentation. Because the non-local weight has higher weight at the edge, the non-local weight fitted image for fitting the foreground and the background calculated by using the non-local weight can better reflect the edge characteristics of the image, and the segmentation precision is further improved. And the scalable normalization method in the application optimizes the driving term. Firstly, the normalized driving term is between (-1, 1), so that adverse effects caused by the excessively high amplitude and the false edge points can be reduced, and the robustness of segmentation is enhanced. Secondly, the threshold value is determined through the statistical distribution of the amplitude of the driving item, so that the point of the driving item exceeding the threshold value is limited to tend to +/-1, and the point lower than the threshold value tends to 0, thereby eliminating irrelevant details in the driving item, enhancing the capability of the driving item for distinguishing the target from the background, and further improving the segmentation precision.
S3, performing iterative updating on the initial level set function by using a non-local characteristic driving term to obtain a final contour line;
the method comprises the following specific steps:
setting a fitting dirac functionThe fitted dirac function +.>Expressed as:
(14);
wherein,representing the calculated variable fitting the dirac function,/->Is a constant, in this embodiment, < +.>。
Using a non-local feature driving term to carry out iterative updating on the initial level set function, wherein a level set iterative updating formula is expressed as follows:
(15);
wherein,the iteration times; />For normalization function->Is a constant; />Representing the initial level set iteration +.>Post-level set function ++>Representing the initial level set iteration +.>A level set function after the secondary; />To->Is an average filter of a convolution kernel template, +.>Is +.>,/>Is a constant, in this embodiment, < +.>The value range of (5) is [5,17 ]];/>Is a convolution operation symbol; />Is the time step; in this embodiment, <' > a->,/>。
The termination conditions are set according to the following formula:
(16);
wherein,for terminating the condition coefficient, in this embodiment, < +.>。
When the initial level set function is iteratively updated by using the non-local feature driving term, stopping the iterative updating of the level set function when the termination condition is met, and enabling the level set function to be updated lastAs a final contour.
The method and the device accurately extract image characteristics and reflect target boundary characteristics by using the non-local characteristic driving terms established by combining the non-local characteristic images and the scalable normalization method, and iteratively update the initial level set function so as to obtain a segmentation result. The non-local feature driving item does not participate in the level set iteration updating process, but is obtained by only being calculated once before the level set iteration, so that the segmentation efficiency is improved. Because the non-local characteristic driving term is pre-calculated before level set iteration, the influence of the initial contour line on the segmentation accuracy is small, and the segmentation robustness is enhanced.
And S4, segmenting the image to be segmented by using the final contour line to obtain an image segmentation result.
In this embodiment, the method and the existing active contour model described in the present application are used to perform an image segmentation simulation experiment, and the experimental results are compared. Existing active contour models include: LGJDACM (local and global region based on Jeffreys divergence active contour model, jeffreys divergence local and global region based active contour model), APFJDACM (adaptive pre-fitting function Jeffreys divergence active contour model, adaptive pre-fit function Jeffreys divergence active contour model), OLPFIACM (optimized local pre-fitting image active contour model, optimized local pre-fit image active contour model), LKLDACM (local Kullback-Leibler divergence active contour model ).
All experiments were performed in MatlabR2019a on a 1.8GHz intel borui 5 personal computer. Images used for segmentation experiments and comparison experiments were from the BSDS gallery and the Weizmann gallery. The color image is converted into a gray scale image before segmentation. In all experiments, the rectangular solid line represents the initial contour line and the dashed line represents the final segmentation curve.
The parameters are set as follows:,/>,/>,/>,/>,/>,/>,/>,/>,/>and setting the maximum iteration number N and an initial contour line.
First, 4 natural images are selected from the BSDS gallery and the Weizmann gallery, and the four images are segmented by the method described in the application. As shown in fig. 2, each line from top to bottom is a processing diagram of the 4 images, respectively. The first column is a schematic diagram of an original image to be segmented and an initial contour, the second column is a schematic diagram of a driving item combined with non-local features only, the third column is a schematic diagram of a non-local feature driving item combined with the non-local features and a scalable normalization method, and the fourth column is a final segmentation result obtained by the method. Experimental results show that the driving item combined with the non-local features can reflect edge features of the image to be segmented, but irrelevant details except for the target edge are mixed, and the non-local feature driving item after the scalable normalization method can eliminate irrelevant details in the driving item while retaining the target edge, so that the capability of the driving item in distinguishing the target and the background is enhanced. Therefore, the non-local feature driving term can accurately extract the target edge feature of the image, so that an accurate segmentation result can be obtained.
Next, 8 color images were selected from the BSDS gallery and Weizmann gallery, and the 8 images were segmented using the five image segmentation models LGJDACM, APFJDACM, OLPFIACM, LKLDACM and the methods described herein, respectively. As shown in FIG. 3, each row of images is numbered 1-8 in sequence from top to bottom, wherein the images numbered 6-8 are artificially added with multiplicative noise with the mean value of 0 and the variance of 0.05. The first column in fig. 3 is the original image to be segmented and the initial contour, and the second column to the sixth column are the segmentation results of LGJDACM, APFJDACM, OLPFIACM, LKLDACM and the method described herein, respectively. As can be seen from fig. 3, the present application can obtain accurate segmentation results for both noiseless images and noisy images, and the segmentation results of the present application are superior to LGJDACM, OLPFIACM, LKLDACM and approach APFJDACM from the perspective of overall image segmentation results.
To further compare the performance of the five image segmentation models, comparisons are made in terms of segmentation efficiency and segmentation accuracy, respectively. As shown in table 1, table 1 is a comparison table of the running times of the five image segmentation models, and the shorter the running time, the higher the segmentation efficiency. In terms of the segmentation accuracy, two indexes, namely DSC (Dice Similarity Coefficient ) and IoU (Intersection of Union, overlapping degree), are adopted to measure the segmentation accuracy of each model.
DSC definition is(17);
IoU is defined as(18);
Wherein,is the standard target region provided in BSDS gallery and Weizmann gallery,/and->Is the experimentally obtained target region. The closer the values of DSC and IoU are to 1, the higher the segmentation accuracy is.
As shown in table 2, table 2 is a comparative table of DSC and IoU for these five image segmentation models. As can be seen from the data in tables 1 and 2, compared with LGJDACM, APFJDACM, OLPFIACM and LKLDACM, the method can quickly and accurately divide the noiseless and noisy images, and has certain advantages in the division efficiency and the division precision.
Table 1 run time comparison table (seconds) for five image segmentation models
TABLE 2 DSC and IoU comparison tables of five image segmentation models (DSC/IoU)
In a second aspect, as shown in fig. 4, the present embodiment provides an image segmentation system 600 based on non-local features, including: an initial contour setting module 601, a non-local feature driving term establishing module 602, a final contour generating module 603, and an image dividing module 604;
the initial contour line setting module 601 is configured to obtain an image to be segmented, and set an initial contour line in the image to be segmented by using a level set method;
the non-local feature driving term establishing module 602 is configured to establish a non-local feature driving term by combining the non-local feature image and the scalable normalization method;
the final contour generating module 603 is configured to iteratively update the initial level set function using the non-local feature driving term to obtain a final contour;
the image segmentation module 604 is configured to segment the image to be segmented using the final contour line, and obtain an image segmentation result.
According to the non-local feature-based image segmentation system provided in the present embodiment, since the non-local feature-based image segmentation system is used for implementing the steps of the non-local feature-based image segmentation method provided in the first aspect of the present invention, the non-local feature-based image segmentation system has all the technical effects of the non-local feature-based image segmentation method, which are not described herein.
In a third aspect, an embodiment of the present invention provides a computer readable storage medium having stored thereon a computer program, which when executed implements the non-local feature based image segmentation method according to any one of the first aspects above.
In a fourth aspect, an embodiment of the present invention provides a storage device, including a storage medium and a processor, where the storage medium stores a computer program, where the program when executed by the processor implements the non-local feature-based image segmentation method according to any one of the first aspects.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, the present invention should also include such modifications and variations provided that they come within the scope of the following claims and their equivalents.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that alterations, modifications, substitutions and variations may be made in the above embodiments by those skilled in the art within the scope of the invention.
Claims (10)
1. A method of image segmentation based on non-local features, the method comprising:
acquiring an image to be segmented, and setting an initial contour line in the image to be segmented by using a level set method;
establishing a non-local feature driving item by combining the non-local feature image and a scalable normalization method;
iteratively updating the initial level set function by using the non-local characteristic driving term to obtain a final contour line;
and segmenting the image to be segmented by using the final contour line to obtain an image segmentation result.
2. The non-local feature-based image segmentation method according to claim 1, wherein an initial contour line is set in the image to be segmented using a level set method, specifically as follows:
representing the initial contour line by using the initial level set function, wherein when the vector pixel point of the image to be segmented is outside the initial contour line, the value of the initial level set function is 1; when the vector pixel point of the image to be segmented is in the initial contour line or on the initial contour line, the value of the initial level set function is-1.
3. The non-local feature-based image segmentation method as set forth in claim 1, wherein the method of creating the non-local feature driving term is expressed as:
;
wherein,for vector pixel points in the image to be segmented, < > in->Driving an item for said non-local feature, +.>For determining the direction sign of the curve evolution direction +.>To adjust the constant of the amplitude of the non-local feature driving term +.>For a scalable normalization function, +.>Is a telescopic factor->Gray values of the image that are non-local feature differences;
the direction symbol of the evolution direction of the judgment curve is expressed as follows:
;
wherein,representing the initial level set function, +.>Representing an average calculation function, +.>Average gray value of vector pixel points representing the value of the initial level set function less than 0,/v>Representing the average gray value of the vector pixel points with the value of the initial level set function being more than or equal to 0;
the scalable normalization function is expressed as:
;
wherein,a calculation variable representing said scalable normalization function, -a calculation variable representing said scalable normalization function>Is a telescopic factor->Is->The abscissa of the positive turning point, +.>To distinguish between the threshold of the edge and non-edge regions of the drive term.
4. A non-local feature based image segmentation method according to claim 3, wherein the gray values of the non-local feature difference image are expressed as:
;
wherein,for images processed by the non-local mean algorithm, < >>Non-local weighted fitting images for the image to be segmented for fitting foreground and background,/for the image to be segmented>As a logarithmic function>The standard deviation of the gray scale of the image to be segmented is obtained;
the non-local mean algorithm processed image is expressed as:
;
wherein,to->Is the center and the side center distance is +.>Square window of>,/>For smooth parameters +.>To lower round symbol->For window->Vector pixel points of (2); />For the gray value of the image to be segmented, < >>Is->And->Non-local weights in between.
5. The non-local feature based image segmentation method as set forth in claim 4, wherein the non-local weighted fit image comprises:
a non-local weighted fit image of the high gray sub-region, expressed as:
;
a non-local weighted fit image of the low gray sub-region, expressed as:
;
wherein,for sub-areas where the grey value of the image is higher than the average grey value,/->For sub-areas where the grey value of the image is lower than the average grey value,/->For window->Average gray values of all vector pixels.
6. The non-local feature based image segmentation method as set forth in claim 5, wherein the iterative updating is performed using the following formula:
;
wherein,for the number of iterations->For normalization function->Is a constant, +.>Representing the initial level set iteration +.>Post-level set function ++>Representing the initial level set iteration +.>Post-level set function ++>To->Is an average filter of a convolution kernel template, +.>Is +.>,/>Is a constant, +.>For convolution operation symbol>For the time step size of the time step,to fit the dirac function, +.>Is a constant.
7. The method of image segmentation based on non-local features according to claim 6, wherein in the step of iteratively updating the initial level set function using the non-local feature driving term, when a termination condition is satisfiedWhen the level set function stops iterative updating, and the last updated level set function is used as the final contour line, wherein +_>Is a termination condition coefficient.
8. An image segmentation system based on non-local features, the system comprising:
the device comprises an initial contour line setting module, a contour line segmentation module and a contour line segmentation module, wherein the initial contour line setting module is used for acquiring an image to be segmented and setting an initial contour line in the image to be segmented by using a level set method;
the non-local feature driving item establishing module is used for establishing a non-local feature driving item by combining the non-local feature image and a scalable normalization method;
the final contour line generation module is used for carrying out iterative updating on the initial level set function by using the non-local characteristic driving term to obtain a final contour line;
and the image segmentation module is used for segmenting the image to be segmented by using the final contour line to obtain an image segmentation result.
9. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the non-local feature based image segmentation method according to any one of claims 1 to 7.
10. A storage device comprising a storage medium storing a computer program, and a processor, wherein the processor implements the non-local feature based image segmentation method of any one of claims 1-7 when executing the computer program.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410079491.7A CN117593323B (en) | 2024-01-19 | 2024-01-19 | Image segmentation method, system, medium and device based on non-local features |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410079491.7A CN117593323B (en) | 2024-01-19 | 2024-01-19 | Image segmentation method, system, medium and device based on non-local features |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117593323A true CN117593323A (en) | 2024-02-23 |
CN117593323B CN117593323B (en) | 2024-03-29 |
Family
ID=89922389
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410079491.7A Active CN117593323B (en) | 2024-01-19 | 2024-01-19 | Image segmentation method, system, medium and device based on non-local features |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117593323B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117994277A (en) * | 2024-04-02 | 2024-05-07 | 苏州大学 | Target segmentation method, device, medium and system for non-uniform illumination image |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111462145A (en) * | 2020-04-01 | 2020-07-28 | 重庆大学 | Active contour image segmentation method based on double-weight symbol pressure function |
CN115984312A (en) * | 2023-03-15 | 2023-04-18 | 苏州大学 | Image segmentation method, electronic device and computer-readable storage medium |
CN117173201A (en) * | 2023-11-01 | 2023-12-05 | 苏州大学 | Second order differential image segmentation method, system, medium and device |
-
2024
- 2024-01-19 CN CN202410079491.7A patent/CN117593323B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111462145A (en) * | 2020-04-01 | 2020-07-28 | 重庆大学 | Active contour image segmentation method based on double-weight symbol pressure function |
CN115984312A (en) * | 2023-03-15 | 2023-04-18 | 苏州大学 | Image segmentation method, electronic device and computer-readable storage medium |
CN117173201A (en) * | 2023-11-01 | 2023-12-05 | 苏州大学 | Second order differential image segmentation method, system, medium and device |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117994277A (en) * | 2024-04-02 | 2024-05-07 | 苏州大学 | Target segmentation method, device, medium and system for non-uniform illumination image |
CN117994277B (en) * | 2024-04-02 | 2024-06-07 | 苏州大学 | Target segmentation method, device, medium and system for non-uniform illumination image |
Also Published As
Publication number | Publication date |
---|---|
CN117593323B (en) | 2024-03-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109086714B (en) | Form recognition method, recognition system and computer device | |
CN110335290B (en) | Twin candidate region generation network target tracking method based on attention mechanism | |
US20180114313A1 (en) | Medical Image Segmentation Method and Apparatus | |
CN117593323B (en) | Image segmentation method, system, medium and device based on non-local features | |
CN112819772B (en) | High-precision rapid pattern detection and recognition method | |
CN110689545A (en) | Fuzzy boundary image automatic segmentation method based on active contour and deep learning | |
CN109767439A (en) | A kind of multiple dimensioned difference of self-adapting window and the object detection method of bilateral filtering | |
CN108520214A (en) | A kind of finger vein identification method based on multiple dimensioned HOG and SVM | |
WO2021136368A1 (en) | Method and apparatus for automatically detecting pectoralis major region in molybdenum target image | |
CN111583279A (en) | Super-pixel image segmentation method based on PCBA | |
CN105404868B (en) | The rapid detection method of text in a kind of complex background based on interaction platform | |
CN105046278B (en) | The optimization method of Adaboost detection algorithm based on Haar feature | |
WO2010043954A1 (en) | Method, apparatus and computer program product for providing pattern detection with unknown noise levels | |
CN110991547A (en) | Image significance detection method based on multi-feature optimal fusion | |
CN113392854A (en) | Image texture feature extraction and classification method | |
CN111310609A (en) | Video target detection method based on time sequence information and local feature similarity | |
CN107564013B (en) | Scene segmentation correction method and system fusing local information | |
CN112614170A (en) | Fourier power spectrum-based single particle image registration method for cryoelectron microscope | |
Xu et al. | Improved Canny Edge Detection Operator | |
CN110176021B (en) | Level set image segmentation method and system for saliency information combined with brightness correction | |
CN112200730B (en) | Image filtering processing method, device, equipment and storage medium | |
CN112419407B (en) | Cloud cluster displacement vector calculation method and device based on cloud cluster edge recognition | |
CN113033256B (en) | Training method and device for fingertip detection model | |
CN109934190B (en) | Self-adaptive highlight face image texture recovery method based on deformed Gaussian kernel function | |
CN114529518A (en) | Image pyramid and NLM-based image enhancement method for cryoelectron microscope |
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 |