CN101408979A - Method for processing human jaw facial bone CT image digitalization based on inheritance algorithm - Google Patents
Method for processing human jaw facial bone CT image digitalization based on inheritance algorithm Download PDFInfo
- Publication number
- CN101408979A CN101408979A CNA2008102028487A CN200810202848A CN101408979A CN 101408979 A CN101408979 A CN 101408979A CN A2008102028487 A CNA2008102028487 A CN A2008102028487A CN 200810202848 A CN200810202848 A CN 200810202848A CN 101408979 A CN101408979 A CN 101408979A
- Authority
- CN
- China
- Prior art keywords
- genetic algorithm
- image
- facial bone
- processing
- human jaw
- 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.)
- Pending
Links
Images
Landscapes
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a human body jaw facial bone CT image digitization processing method based on genetic algorithm and relates to the image processing technical field; the method solves the technical problem that the digitization processing of medical images is easily operated; the digitization processing method comprises the following steps: 1) gray value quantification is carried out on the human body jaw facial bone CT image, and the range is (0, 1), thus obtaining grey level histogram of the image; 2) the calculation of genetic algorithm is carried out to obtain optimal threshold; 3) binarization digitization processing is carried out on the CT image according to the optimal threshold; 4) mathematical morphology processing is carried out to obtain boundary curve of binary image. The digitization processing method of the invention has the characteristics of the digitization processing of being easily to be operated and being capable of realizing the medical images.
Description
Technical field
The present invention relates to image processing techniques, particularly relate to a kind of method for processing human jaw facial bone CT image digitalization based on genetic algorithm.
Background technology
Along with developing rapidly of computer technology and advanced medicine equipment, equipment, some advanced persons' such as X-X-ray machine X, CT, nuclear magnetic resonance, DSA detection and the attention that diagnostic means more and more is subjected to numerous medical workers and be extensive use of.The medical image of magnanimity and the problem of image and storage and inquiry have also been produced thereupon.Therefore the digitized processing and the storage of medical image, image become problem that needs to be resolved hurrily on the clinical medicine.
A basic problem during image digitization is handled is Threshold Segmentation (Thresholding).Its objective is image is divided into respective regions to make things convenient for subsequent processes by different gray-scale values.Selection of threshold is the prerequisite of Threshold Segmentation.Selection of threshold is improper, can influence the shape size of target, even target is lost.Selection of threshold is the problem of optimizing in view data just.Because the digital image information amount is big, if can not utilize relevant knowledge to dwindle the search volume, then may produce the shot array of search.In recent years, Chinese scholars has been carried out research extensively and profoundly at this problem, has proposed multiple selection of threshold method, but these methods exist in varying degrees carry out efficient low, be easy to be absorbed in problem such as locally optimal solution.Therefore, the selection of threshold method of seeking a kind of globally optimal solution efficiently is the problem that people paid close attention to always.Genetic algorithm (Genetic Algorithm, be called for short GA) is especially effective one of algorithm wherein.
Genetic algorithm is at first proposed in 1975 by American scholar Holand.It has embodied life science and engineering science intersects mutually, infiltration and promotion.Genetic algorithm is subjected to the enlightenment of nature biotechnology evolutionary process, uses for reference natural selection and hereditary naturally mechanism, can obtain automatically in search procedure and the knowledge that accumulates relevant search volume, and control search procedure adaptively in the hope of optimum solution.Genetic algorithm is a kind of overall multiple spot searching algorithm at random, and its two big principal features are that the information between colony's search strategy and the individual in population exchanges mutually.Genetic algorithm is from arbitrary initialized colony, by selecting at random, intersect and genetic manipulation such as variation, evolves to the zone of becoming better and better in the search volume, until reaching optimum solution point with making colony's generation generation.The major advantage of GA be simple, general, the robust type is strong, be applicable to parallel processing, thereby be used widely.
Genetic manipulation is the most important operation of GA.Selecting, intersecting and make a variation is 3 main operation operators of genetic algorithm.
The purpose of selection operation is to select defect individual from current colony, and they are had an opportunity as parent.Select according to being individual fitness value.The ideal adaptation degree is high more, and its selecteed probability is just big more.The selection strategy that generally adopts all is to select with the proportional probability of fitness at present.Such as roulette wheel strategy, best retention strategy, ordering strategy etc.
Interlace operation is a topmost genetic manipulation in the genetic algorithm.Select individuality at random as parent by certain probability in current colony, parent produces a new generation through hybridization.New individuality has kept parents' portion gene, has introduced new gene again.
Mutation operation is very delicate genetic manipulation, properly cooperate with interlace operation, and purpose is to excavate the diversity of individual in population, overcomes the disadvantage that might produce local solution.
Summary of the invention
At the defective that exists in the above-mentioned prior art, technical matters to be solved by this invention provides a kind of easy operating, can realize the method for processing human jaw facial bone CT image digitalization based on genetic algorithm of the digitized processing of medical image.
In order to solve the problems of the technologies described above, a kind of method for processing human jaw facial bone CT image digitalization based on genetic algorithm provided by the present invention is characterized in that the step of digitalized processing method comprises:
1) gray-scale value with the human jaw facial bone CT image quantizes, and its scope is [0,1]; Can obtain its grey level histogram;
2) through the calculating of genetic algorithm, obtain optimal threshold;
3) by this optimal threshold the CT image is carried out the binaryzation digitized processing;
4) handle through mathematical morphology, obtain boundary curve of binary image.
Further, described step 2) in, the selection operator in the genetic algorithm adopts the fitness ratio method; Crossover operator adopts the average intersection in the protruding intersection; Mutation operator adopts dynamic variation.
Further, described step 2) in, the major parameter population size of genetic algorithm is 10-160.
Further, described step 2) in, the crossover probability in the genetic manipulation of genetic algorithm is 0.25-1.00.
Further, described step 2) in, the variation probability in the genetic manipulation of genetic algorithm is 0.05-0.3.
Further, described step 2) in, when handling mandibular CT image, on the basis of the threshold value t that GA determines, between [t-0.1,1], carry out a maximum between-cluster variance again and calculate, to obtain optimal threshold.
Further, in the described step 4), at last boundary curve is carried out fairing processing.
Utilize the method for processing human jaw facial bone CT image digitalization based on genetic algorithm provided by the invention, owing in genetic manipulation, take real coding to replace the binary string coding techniques, avoiding that the gradation of image value is converted to binary string operates again, thereby shortening code length, the expanded search space, and meet thinking habit, easy operating has obtained the optimal threshold of human jaw facial bone CT image slices.On this threshold basis, obtain its boundary curve, for the digitized processing of medical image lays the first stone.
Description of drawings
Fig. 1 is the CT image of human lower jawbone tomography;
Fig. 2 is the grey level histogram of CT image among Fig. 1;
Fig. 3 is a mandibular CT image binary map among Fig. 1;
Fig. 4 is a boundary curve of binary image among Fig. 3;
Fig. 5 is the CT image of another tomography of human lower jawbone;
Fig. 6 (a) is the CT image binary map of mandibular among Fig. 5;
Fig. 6 (b) is the boundary curve that obtains through genetic manipulation among Fig. 5.
Embodiment
Below in conjunction with description of drawings embodiments of the invention are described in further detail, but present embodiment is not limited to the present invention, every employing similarity method of the present invention and similar variation thereof all should be listed protection scope of the present invention in.
A kind of method for processing human jaw facial bone CT image digitalization based on genetic algorithm that the embodiment of the invention provided is characterized in, the step of digitalized processing method comprises:
1) gray-scale value with the human jaw facial bone CT image quantizes, and its scope is [0,1]; Can obtain its grey level histogram;
2) through the calculating of genetic algorithm, obtain optimal threshold; And in genetic manipulation, take real coding to replace the binary string coding techniques, avoiding that the gradation of image value is converted to binary string operates again, thereby shortening code length, the expanded search space, and meet thinking habit, easy operating has obtained the optimal threshold of human jaw facial bone CT image slices;
3) by this threshold value the CT image is carried out the binaryzation digitized processing;
4) handle through mathematical morphology, obtain boundary curve of binary image;
5) boundary curve is carried out fairing processing.
Comprise five fundamentals in genetic algorithm (GA) processing procedure, i.e. coding, initial population setting, fitness function, genetic manipulation and GA controlled variable.Here adopt real coding to replace binary string chromosome, so that the character of more direct description problem.
3 operation operator settings of genetic algorithm:
Genetic algorithm combines the advantage of beam search and random search in the process of search optimum solution.In the incipient stage of genetic search, crossing operation is because the initial population that produces has diversity and trends towards in extensive search at random.Along with height just when the acquisition of separating, crossing operation trend towards these separate around the search.
Select operator to adopt fitness ratio method (fitness proportional model), claim roulette wheel method or Monte Carlo method (Monte Carlo) to select again.Calculate earlier each chromosome separately just when and all chromosomes just when summation.At equally distributed pseudo random number r of [0,1] interval generation.Since No. 1 chromosome, each is chromosomal just when addition successively, when adding up and during more than or equal to r, and the individuality of that chromosome of Jia Ruing for selecting at last.
Crossover operator adopts the average intersection in the protruding intersection.The i.e. individual x of two parents
1, x
2Do following combination, produce filial generation,
x′
1＝λ
1x
1+λ
2x
2
X '
2=λ
1x
2+ λ
2x
1, λ wherein
1=λ
2=0.5 (1)
The offspring that protruding intersection produces is positioned on the real segment at two-dimensional space.
Mutation operator adopts dynamic variation, and it is for improving precision, increasing the fine tuning ability and design.For parent X, if element x
kBe selected and make a variation, then offspring X '=[x
1..., x '
k..., x
n], x ' wherein
kMay select by following two kinds:
Or
X ' wherein
kBe [x
k L, x
k U] in an equally distributed random value.x
k LAnd x
l UUsually can be taken as variable x
kBound, generally can determine by constrained domain.
(g y) returns a value in [0, y] to the function Δ, makes that (g y) is tending towards 0 (g is a genetic algebra) with the g increase to Δ.When this character made primary iteration, search was evenly distributed on whole space, and the later stage then is distributed in the subrange.Δ (g, y) form is:
Wherein r is the random number in [0,1], and G is maximum algebraically, and b is a parameter of determining unevenness.Through repeatedly calculating relatively, it is 2 that this paper chooses b.
The major parameter of control GA treatment effect is population size and genetic manipulation probability.
Population size influences the net result of genetic optimization and the execution efficient of genetic algorithm.Population size is big, can keep individual diversity preventing to be absorbed in local solution, but can increase calculated amount, also may influence individual competition.And the too little optimization performance that has limited genetic algorithm of scale.Population size is generally got between the 10-160.From gray-scale value greater than randomly drawing 10 individualities 0.6 the pixel as initial population.
Crossover probability in the genetic manipulation and the setting of variation probability:
The usage frequency of crossover probability control interlace operation.High crossing-over rate means individual the renewal soon, reaches bigger solution space, has reduced the probability of obtaining non-optimum solution, but the destroyed possibility increase of high performance mode, and search for unnecessary solution space and want consumes resources; Crossing-over rate is low excessively, and the search meeting is because the scope of detecting is little and blunt.Crossing-over rate is generally got between the 0.25-1.00.
Colony's diversity is kept in variation, the gene that does not have in the initial population is provided, or gives the gene of losing in the search for change.Too low aberration rate may make some desirable genes can not enter selection; Too high aberration rate then makes search be tending towards random search, and the good characteristic of parent is not inherited in filial generation, and algorithm loses self-learning capability.Aberration rate is generally got between the 0.05-0.3.
At the characteristics of this research, and through getting the comparison that experimentizes of different numerical value, it is 0.3 that this paper chooses crossing-over rate, and aberration rate is 0.1.Shutting down one of criterion is that maximum iteration time G=20 is (after iteration 20 times; reached desirable convergence result); two of criterion is ratio ranges (adjacent generations ratio is tending towards convergence in this interval) between [0.995,1.005] of the average fitness value of average fitness value of current colony and previous generation.
The further optimization that quasi-optimal is separated---maximum between-cluster variance calculates:
In Flame Image Process, separating of GA optimizing may be optimum solution, also may be that quasi-optimal is separated.When handling mandibular CT image, on the basis of the threshold value t that GA determines, between [t-0.1,1], carry out a maximum between-cluster variance again and calculate, to obtain optimal threshold.
Maximum variance between clusters is divided into two class C to the pixel in the image by gray level with threshold value t
0And C
1, C
0By gray-scale value [0, t] pixel form C
1By gray-scale value [t, g-1] pixel form (g is a gradation of image progression).Inter-class variance σ (t)
2Expression,
σ(t)
2＝n
1(t)×n
2(t)×[p
1(t)-p
2(t)]
2 (4)
In the formula: n
1(t) be C
0In the number of pixels that comprises;
n
2(t) be C
1In the number of pixels that comprises;
p
1(t) be C
0In the average gray value of the pixel that comprises;
p
2(t) be C
1In the average gray value of the pixel that comprises;
T from 0 to 1 value, the t when making that σ is maximum is optimal threshold t
*, promptly
Optimal threshold
Fig. 1 is the CT image of a width of cloth with a tomography of the human lower jawbone of digital camera acquisition.From Fig. 1, can be observed, brightness maximum (being the gray-scale value maximum) be mandibular.Border between soft tissue around it and its is not fairly obvious, has the zone of transition that the gray-value variation amplitude is mild.Mandibular and background area will be separated, must determine its boundary curve scientifically, exactly, this at first will select one appropriate " threshold value ".Quantize with the gray-scale value of MATLAB program with Fig. 1, its scope is [0,1].Can obtain its grey level histogram, as Fig. 2.In Fig. 2, gray-scale value is the pixel of mandibular near 1 pixel.Obviously, threshold value should be in the crossing part near 1 peak and valley.Can adopt genetic algorithm to determine this threshold value.
Through the calculating of genetic algorithm, obtaining optimal threshold is 0.9686.In Fig. 2, the crossing part of the peak and valley of this threshold value close 1.The CT image of mandibular among Fig. 1 is carried out result such as Fig. 3 of binaryzation by this threshold value:
Handle through mathematical morphology again, obtain boundary curve of binary image such as Fig. 4 among Fig. 3.Compare with mandibular in the CT image (Fig. 1), the result is more satisfactory.After obtaining the border, just can further carry out fairing processing to boundary curve.
Use the same method and handle CT image such as Fig. 5 of another width of cloth human lower jawbone tomography, obtaining threshold value is 0.9176, and its binary map and boundary curve are shown in Fig. 6 (a) and (b).
Claims (9)
1, a kind of method for processing human jaw facial bone CT image digitalization based on genetic algorithm is characterized in that, the step of digitalized processing method comprises:
1) gray-scale value with the human jaw facial bone CT image quantizes, and its scope is [0,1]; Can obtain its grey level histogram;
2) through the calculating of genetic algorithm, obtain optimal threshold;
3) by this optimal threshold the CT image is carried out the binaryzation digitized processing;
4) handle through mathematical morphology, obtain boundary curve of binary image.
2, the method for processing human jaw facial bone CT image digitalization based on genetic algorithm according to claim 1 is characterized in that, described step 2) in, the selection operator in the genetic algorithm adopts the fitness ratio method.
3, the method for processing human jaw facial bone CT image digitalization based on genetic algorithm according to claim 1 is characterized in that, described step 2) in, the crossover operator in the genetic algorithm adopts the average intersection in the protruding intersection.
4, the method for processing human jaw facial bone CT image digitalization based on genetic algorithm according to claim 1 is characterized in that, described step 2) in, the mutation operator in the genetic algorithm adopts dynamic variation.
5, the method for processing human jaw facial bone CT image digitalization based on genetic algorithm according to claim 1 is characterized in that, described step 2) in, the major parameter population size of genetic algorithm is 10-160.
6, the method for processing human jaw facial bone CT image digitalization based on genetic algorithm according to claim 1 is characterized in that, described step 2) in, the crossover probability in the genetic manipulation of genetic algorithm is 0.25-1.00.
7, the method for processing human jaw facial bone CT image digitalization based on genetic algorithm according to claim 1 is characterized in that, described step 2) in, the variation probability in the genetic manipulation of genetic algorithm is 0.05-0.3.
8, the method for processing human jaw facial bone CT image digitalization based on genetic algorithm according to claim 1, it is characterized in that, described step 2) in, on the basis of the threshold value t that genetic algorithm is determined, again at [t-0.1,1] carries out a maximum between-cluster variance between and calculate, to obtain optimal threshold.
9, the method for processing human jaw facial bone CT image digitalization based on genetic algorithm according to claim 1 is characterized in that, in the described step 4), at last boundary curve is carried out fairing processing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008102028487A CN101408979A (en) | 2008-11-18 | 2008-11-18 | Method for processing human jaw facial bone CT image digitalization based on inheritance algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008102028487A CN101408979A (en) | 2008-11-18 | 2008-11-18 | Method for processing human jaw facial bone CT image digitalization based on inheritance algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101408979A true CN101408979A (en) | 2009-04-15 |
Family
ID=40571980
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2008102028487A Pending CN101408979A (en) | 2008-11-18 | 2008-11-18 | Method for processing human jaw facial bone CT image digitalization based on inheritance algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101408979A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103390284A (en) * | 2012-05-08 | 2013-11-13 | 西门子公司 | CT image reconstruction in the extended field of view |
| CN105631876A (en) * | 2015-12-29 | 2016-06-01 | 中国兵器科学研究院宁波分院 | CT image resolution automatic test method based on global binarization |
| CN108858198A (en) * | 2018-07-26 | 2018-11-23 | 西北工业大学 | A kind of robotic arm path planing method based on Genetic Simulated Annealing Algorithm |
-
2008
- 2008-11-18 CN CNA2008102028487A patent/CN101408979A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103390284A (en) * | 2012-05-08 | 2013-11-13 | 西门子公司 | CT image reconstruction in the extended field of view |
| US9495769B2 (en) | 2012-05-08 | 2016-11-15 | Siemens Aktiengesellschaft | CT image reconstruction in the extended field of view |
| CN103390284B (en) * | 2012-05-08 | 2016-12-28 | 西门子公司 | The CT image reconstruction measured in field in extension |
| CN105631876A (en) * | 2015-12-29 | 2016-06-01 | 中国兵器科学研究院宁波分院 | CT image resolution automatic test method based on global binarization |
| CN108858198A (en) * | 2018-07-26 | 2018-11-23 | 西北工业大学 | A kind of robotic arm path planing method based on Genetic Simulated Annealing Algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tarkhaneh et al. | An adaptive differential evolution algorithm to optimal multi-level thresholding for MRI brain image segmentation | |
| Hill et al. | Model-based image interpretation using genetic algorithms | |
| Fernandes et al. | Self-regulated artificial ant colonies on digital image habitats | |
| CN104134221B (en) | Image segmentation method based on self-adaptive genetic algorithm and OTSU algorithm | |
| Naga Srinivasu et al. | A comparative review of optimisation techniques in segmentation of brain MR images | |
| Shahrezaee | Image segmentation based on world cup optimization algorithm | |
| Karkavitsas et al. | Object localization in medical images using genetic algorithms | |
| Li et al. | An adaptive mutation operator for particle swarm optimization | |
| CN101408979A (en) | Method for processing human jaw facial bone CT image digitalization based on inheritance algorithm | |
| Peng et al. | Memetic differential evolution with an improved contraction criterion | |
| Aguirre et al. | Genetic algorithms on NK-landscapes: Effects of selection, drift, mutation, and recombination | |
| Liu et al. | A novel PSO model based on simulating human social communication behavior | |
| Burdsall et al. | Evolving fuzzy prototypes for efficient data clustering | |
| Hintz-Madsen et al. | Design and evaluation of neural classifiers application to skin lesion classification | |
| Sathya et al. | Comparison of intelligent techniques for multilevel thresholding problem | |
| CN109711460A (en) | The acquisition methods and device of initial cluster center | |
| Xiang et al. | A parallelepiped multispectral image classifier using genetic algorithms | |
| Abedi | Edge detection algorithm for in-pixel lighting via genetic optimization algorithm | |
| Schonfeld et al. | A comparison of the robustness of evolutionary computation and random walks | |
| Ragusa et al. | Augmenting evolution with bio-inspired “super explorers” | |
| CN112115969A (en) | Method and device for optimizing FKNN model parameters based on variant goblet sea squirt group algorithm | |
| Kinzett et al. | Investigation of simplification threshold and noise level of input data in numerical simplification of genetic programs | |
| Bhowan et al. | AUC analysis of the pareto-front using multi-objective GP for classification with unbalanced data | |
| Ashlock et al. | A note on population size inspired by the extinction of mammoths | |
| Chandy et al. | Genetic algorithm based location of optic disc in retinal images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20090415 |