Two-dimensional otsu threshold segmentation method for shield tail gap image based on improved firefly algorithm

Abstract

Disclosed is a shield tail gap image two-dimensional Otsu threshold segmentation method based on an improved firefly algorithm, which comprises the following steps: generating a population randomly in the gray scale [0, 255] range of a pipe segment image; obtaining the optimal two-dimensional firefly position in the population through iteration, which comprises the following steps: calculating the maximum brightness of each firefly in the population; sorting the maximum brightness of the fireflies in the population; for each firefly, forming a local colony with several fireflies whose brightness is lower than that of the firefly; performing an attraction operation between the fireflies in the local colony; updating the position of the firefly, and introducing an attraction force weight to strengthen the connection between the local colonies when updating the position; updating the maximum brightness of the firefly according to the position of the firefly after movement; and outputting the optimal two-dimensional firefly position after multiple iterations, so as to perform threshold segmentation on the pipe segment gray scale image.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of shield construction, and relates to a shield construction monitoring instrument, in particular to a shield tail gap image two-dimensional Otsu threshold segmentation method based on an improved firefly algorithm. BACKGROUND

[0002] With the need for large-scale construction of urban subway tunnels and other underground projects, shield construction technology and equipment have been rapidly developed and applied worldwide. Shield tail gap is a key technical parameter that needs to be accurately measured and real-time monitored during shield tunneling construction, and plays a crucial role in controlling the attitude of the shield machine and ensuring the efficiency and safety of shield construction. Shield tail gap refers to the gap distance between the outer diameter of the segment and the inner diameter of the shield shell. When the shield tail gap changes exceed the set range, excessive extrusion between the shield tail and the segment will occur, accelerating the wear of the shield tail sealing brush, slowing down the tunneling speed, and even causing segment misalignment or damage, leading to serious consequences such as tunnel leakage and ground subsidence. Therefore, it is necessary to measure the shield tail gap in real time, continuously and periodically to ensure the efficiency and safety of shield construction.

[0003] Manual measurement of shield tail gap not only consumes manpower but also has poor accuracy. Some scholars have proposed various shield tail gap measurement methods based on machine vision, which all require real-time segmentation of the segment area to extract the segment region of interest image. The Otsu automatic threshold segmentation method is commonly used for segment area extraction, but the traditional one-dimensional Otsu threshold segmentation method is greatly affected by segment image noise, and the traditional two-dimensional Otsu threshold segmentation method has the disadvantages of high computational complexity, slow running speed, and poor real-time performance due to the use of exhaustive method.

[0004] Firefly algorithm is a newly emerging swarm intelligence optimization algorithm, which is widely used in production scheduling, function optimization, image processing and other optimization problems due to its simple concept, fewer parameters and easy implementation. However, firefly algorithm has the defects of high computational complexity, easy to fall into local optimum and slow convergence speed, so when the classic firefly algorithm is applied to shield tail gap image threshold segmentation in practical application, the effect is not ideal, which limits the application of firefly algorithm in shield tail gap image threshold segmentation. SUMMARY

[0005] The present application provides a shield tail gap image two-dimensional Otsu threshold segmentation method and system based on an improved firefly algorithm, which optimizes the two-dimensional Otsu automatic threshold segmentation process of the segment area by improving the firefly algorithm, reduces the number of calculations of two-dimensional inter-class variance, reduces the computational complexity, and enhances the real-time performance of the shield tail gap monitoring program.

[0006] According to a first aspect of an embodiment of the present application, a two-dimensional Otsu threshold segmentation method for a segment gap image based on an improved firefly algorithm is provided, comprising:

[0007] acquiring a segment image;

[0008] randomly generating N two-dimensional fireflies to form a population within a segment image grayscale range of [0, 255];

[0009] obtaining an optimal two-dimensional firefly position in the population through an iterative method, which comprises: calculating the maximum brightness of each firefly in the population; sorting the maximum brightness of the fireflies in the population, for each firefly, forming a local population with several fireflies having lower brightness than the firefly; performing an attraction operation between the fireflies in the local population, calculating the relative brightness and attraction degree of each firefly with other fireflies in the local population; updating the position of the firefly according to the relative brightness and the attraction degree, introducing an attraction force weight to strengthen the connection between the local populations when updating the position; updating the maximum brightness of the firefly according to the position of the firefly after moving; and outputting the optimal two-dimensional firefly position after multiple iterations;

[0010] taking the mean value of the optimal two-dimensional firefly position as the best threshold, performing threshold segmentation on the segment grayscale image using the best threshold, and obtaining a segment region of interest segmentation result.

[0011] According to a second aspect of an embodiment of the present application, a two-dimensional Otsu threshold segmentation system for a segment gap image based on an improved firefly algorithm is provided, comprising:

[0012] an image acquisition module configured to acquire a segment image;

[0013] a population initialization module configured to randomly generate N two-dimensional fireflies to form a population within a segment image grayscale range of [0, 255];

[0014] a firefly attraction model configured to: calculate the maximum brightness of each firefly in the population; sort the maximum brightness of the fireflies in the population, for each firefly, form a local population with several fireflies having lower brightness than the firefly; perform an attraction operation between the fireflies in the local population, calculate the relative brightness and attraction degree of each firefly with other fireflies in the local population; update the position of the firefly according to the relative brightness and the attraction degree, introduce an attraction force weight to strengthen the connection between the local populations when updating the position; update the maximum brightness of the firefly according to the position of the firefly after moving; and output the optimal two-dimensional firefly position after multiple iterations;

[0015] An image segmentation module is configured to: take the mean of the optimal two-dimensional firefly positions as the optimal threshold value, and perform threshold segmentation on the pipe piece gray-scale image using the optimal threshold value to obtain a pipe piece region of interest segmentation result.

[0016] According to a third aspect of the embodiments of the present application, an apparatus is provided, comprising: a processor; a memory comprising one or more program modules; wherein the one or more program modules are stored in the memory and configured to be executed by the processor, and the one or more program modules comprise instructions for implementing the improved firefly algorithm-based two-dimensional Otsu threshold segmentation method for shield tail gap images

[0017] According to a fourth aspect of the embodiments of the present application, a storage medium is provided for storing non-transitory instructions, which when executed by a processor can implement the improved firefly algorithm-based two-dimensional Otsu threshold segmentation method for shield tail gap images.

[0018] The present application introduces a local attraction model into the traditional firefly algorithm, and improves the attraction force formula, introduces an adaptive attraction force weight into the new attraction force formula, and combines the improved firefly algorithm with the traditional two-dimensional Otsu method to form an automatic shield tail gap image threshold segmentation algorithm. The present application can automatically analyze the optimal segmentation threshold value of the shield tail gap image according to the image returned by the shield tail gap monitoring system. At the same time, the local attraction model effectively reduces the complexity of the firefly algorithm and speeds up the convergence speed of the firefly algorithm. The introduction of the adaptive attraction force weight strengthens the attraction between local colonies and avoids problems such as the firefly falling into a local optimal solution and premature convergence. The improved firefly algorithm-based two-dimensional Otsu threshold segmentation method for shield tail gap images is applied to the shield tail gap monitoring system, so that the monitoring system has higher adaptability and robustness in a complex shield environment, ensures the accuracy of the monitoring result, reduces the calculation amount of the detection system, and improves the operation speed of the monitoring system. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments will be briefly introduced below.

[0020] Figure 1 A flowchart of the improved firefly algorithm-based two-dimensional Otsu threshold segmentation method for shield tail gap images provided by an embodiment of the present application.

[0021] Figure 2 Comparison of the effects of the two-dimensional Otsu threshold segmentation methods optimized by different algorithms on pipe piece region of interest segmentation provided by an embodiment of the present application. DETAILED DESCRIPTION

[0022] The shield tail gap monitoring system based on machine vision and double laser calibration is taken as an example, a zoom industrial camera is used to collect the segment side wall image, an image segmentation algorithm is used to extract the segment region of interest, an image processing algorithm is used to extract the region of interest calibration information, and finally the shield tail gap value is calculated according to the monitoring algorithm, so as to realize the real-time monitoring of the shield tail gap. The image segmentation algorithm for extracting the segment region of interest is improved. The image processing algorithm for extracting the region of interest calibration information and the monitoring algorithm for calculating the shield tail gap value adopt the prior art, and the present application will not be described in detail.

[0023] The improved firefly algorithm is combined with the traditional two-dimensional Otsu threshold segmentation algorithm to form an image segmentation algorithm for extracting the segment region of interest, that is, a two-dimensional Otsu threshold segmentation method for shield tail gap image based on improved firefly algorithm. The principle of the image segmentation algorithm is briefly described as follows: firstly, the firefly position is initialized to uniformly distribute the two-dimensional firefly position in the range of [0, 255]; the two-dimensional inter-class variance is taken as the fitness function value, the fitness function value of each firefly is calculated, and the brightness of each firefly is calculated according to the fitness function value; the improved attraction model is used for attraction operation between fireflies; the improved position updating formula is used for updating the position of fireflies, and the brightness of fireflies is updated according to the moved position; after multiple iterations, the global optimal value and the optimal two-dimensional firefly position are output, and the optimal two-dimensional firefly position is the optimal two-dimensional segmentation threshold of the image, and the mean value of the optimal threshold is taken as the best threshold, and the original segment gray image is segmented by using the best threshold, so as to obtain the segment region of interest segmentation result.

[0024] Step 1, acquiring the segment image.

[0025] Step 2, initializing parameters, including firefly population size N, dimension, light intensity absorption coefficient γ, local colony range k, step factor α, and maximum iteration number T max The firefly dimension is 2, so the two-dimensional segmentation threshold T=(s, t) is obtained, wherein T is the segmentation threshold, and (s, t) is a two-dimensional vector.

[0026] Step 3, initializing the population.

[0027] N two-dimensional firefly individuals are randomly generated in the range of [0, 255], and the maximum brightness I0 of each firefly in the population is calculated.

[0028] The two-dimensional Otsu image segmentation method is to search for the maximum value of two-dimensional inter-class variance, so the target function f is defined as the two-dimensional inter-class variance value J(T), that is:

[0029] f=J(T)=p0(μ0-μT ) · (μ0- μ T ) T + p1(μ1- μ T ) · (μ1- μ T ) T

[0030] Where p0, p1are the two-dimensional pixel probability of the tube image foreground and background, μ0, μ1are the two-dimensional gray mean vector of the tube image foreground and background, μ T is the two-dimensional gray mean vector of the tube image global.

[0031] The value of the objective function f is calculated, which is used as the maximum brightness I0of the firefly.

[0032] Step 4, construction of local attraction model.

[0033] In the process of firefly attraction, a new attraction model is adopted. The brightness of the firefly population with size N is sorted. For each firefly, it is combined with k (much smaller than N) fireflies with lower brightness to form a local colony. The mutual attraction of the entire population is replaced by the internal attraction of multiple local colonies. The following attraction and position update are completed within the local colony, effectively reducing the number of attractions caused by the original algorithm.

[0034] Step 5, population movement.

[0035] For each firefly, calculate its relative brightness I and attraction degree A with other fireflies in the local population. The darker firefly in the local colony will move towards the brightest firefly, and the brightest firefly will randomly fly. Therefore, the position of the brightest firefly after movement needs to be updated.

[0036] The relative brightness formula of the firefly is:

[0037]

[0038] The attraction degree formula of the firefly is:

[0039]

[0040] Where, r ij is the spatial distance between firefly i and j. The maximum brightness I0is the brightness of itself (r = 0), which is positively related to the value of the objective function f. The maximum attraction degree A0is the attraction degree when the light source is at (r = 0).

[0041] Firefly j is attracted by firefly i, and the position is updated by the following formula:

[0042] x j (t+1) = ω(t) x j (t) + Aij (r ij )(x i (t)-x j (t))+αrand()+ω(t)(x best (t)-x j (t))

[0043] where x j (t+1) is the position of firefly j at iteration number t+1; x i (t) is the position of firefly i at iteration number t; x j (t) is the position of firefly j at iteration number t; A ij (r ij ) is the attraction degree between two fireflies; alpha is a step factor, a constant in [0, 1]; rand() is a standard normal distribution (0, 1); omega (t) is an adaptive attraction weight; x best (t) is the optimal individual position in the current local colony. It can be seen that the application strengthens the convergence speed of the firefly algorithm by adding the attraction of the adaptive attraction weight and the local optimal individual in the original position update formula.

[0044] The application introduces an adaptive attraction weight in the position update formula, strengthens the connection between local colonies, and the adaptive attraction weight formula is as follows:

[0045]

[0046] where f(x best (t)) is the global optimal value of the tth iteration, f(x best (t-1)) is the global optimal value of the (t-1)th iteration, f i (t-1) represents the value of firefly x i of the (t-1)th iteration, and f i (t-2) represents the value of firefly x i of the (t-2)th iteration. The adaptive attraction weight reflects the attraction of the optimal value in the population to other individuals, that is, the attraction of the population leader to other individuals.

[0047] Step 6, according to the moved position, updating the brightness of the firefly.

[0048] Step 7, judge whether the maximum iteration number is reached, if yes, proceed to the next step, otherwise, turn to step 4 and add 1 to the iteration number.

[0049] Step 8, after the iteration is completed, output the global optimal value and the optimal two-dimensional firefly position.

[0050] Step 9, the optimal two-dimensional firefly position is the optimal two-dimensional image segmentation threshold, and the mean value thereof is taken as the best threshold. The original pipe piece gray image is threshold segmented using the best threshold, the pixel value of the pixel point greater than the best threshold is set to 255, and the pixel value of the pixel point less than the best threshold is set to 0, so that the pipe piece region of interest segmentation result is obtained.

[0051] In an embodiment, a shield tail gap image two-dimensional Otsu threshold segmentation system based on an improved firefly algorithm is also provided, which comprises an image acquisition module, a population initialization module, a firefly attraction model and an image segmentation module.

[0052] The image acquisition module is configured to acquire a pipe piece image. The population initialization module is configured to randomly generate a population of N two-dimensional fireflies in the gray scale [0, 255] range of the pipe piece image. The firefly attraction model is configured to: calculate the maximum brightness of each firefly in the population; sort the maximum brightness of the fireflies in the population, and for each firefly, form a local colony with several fireflies having lower brightness than it; perform attraction operation between the fireflies in the local colony, and calculate the relative brightness and attraction degree of each firefly with other fireflies in the local colony; update the position of the firefly according to the relative brightness and the attraction degree, and introduce an attraction force weight in the position update to strengthen the connection between the local colonies; update the maximum brightness of the firefly according to the position of the firefly after moving; and output the optimal two-dimensional firefly position after multiple iterations. The image segmentation module is configured to: take the mean value of the optimal two-dimensional firefly position as the best threshold, and perform threshold segmentation on the pipe piece gray image using the best threshold, so as to obtain the pipe piece region of interest segmentation result.

[0053] The firefly attraction model is more detailed in the method described above.

[0054] In some embodiments, an apparatus is also provided. The apparatus can be an image processing device integrated on a shield tunneling machine to calculate the shield tail gap, and of course can also be a personal computer. The apparatus comprises a processor and a memory. The memory is used to store non-transitory instructions (such as one or more program modules). The processor is used to run the non-transitory instructions, and the non-transitory instructions can execute one or more steps of the shield tail gap image two-dimensional Otsu threshold segmentation method based on the improved firefly algorithm described above when executed by the processor. The memory and the processor can be interconnected through a bus system and / or other forms of connection mechanism.

[0055] For example, the processor can be a central processing unit (CPU), a graphics processing unit (GPU), or other form of processing unit having data processing and / or program execution capabilities. For example, the central processing unit (CPU) can be of X86 or ARM architecture, etc. The processor can be a general purpose processor or a special purpose processor, and can control other components in the electronic device to perform desired functions.

[0056] For example, the memory can be a volatile memory and / or a non-volatile memory. The volatile memory may, for example, include a random access memory (RAM), a cache, etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, an erasable programmable read-only memory (EPROM), a compact disk read-only memory (CD-ROM), a USB memory, a flash memory, etc. One or more program modules can be stored on the memory, and the processor can run the one or more program modules to implement various functions of the device.

[0057] In an embodiment, a storage medium for storing non-transitory instructions is also provided, when the non-transitory instructions are executed by an electronic device, one or more steps of the improved firefly algorithm-based two-dimensional Otsu threshold segmentation method for scute tail gap image described above can be implemented. That is, when the improved firefly algorithm-based two-dimensional Otsu threshold segmentation method for scute tail gap image provided by the embodiments of the present application is implemented in the form of software and sold or used as an independent product, it can be stored in a storage medium. For related descriptions of the storage medium, please refer to the corresponding description of the memory of the electronic device above, which will not be repeated here.

[0058] In an embodiment, the firefly population size N = 18, the dimension 2, the light intensity absorption coefficient γ = 0.2, the local colony range k = 4, the step factor α = 0.8, and the maximum number of iterations T max = 20 are set. For each firefly, it can be combined with the nearby 4 fireflies into a local colony structure. After 20 iterations, the optimal firefly position is [132, 154], and the mean value (143) of the optimal two-dimensional segmentation threshold is taken as the best threshold.

[0059] With the above embodiments as the background, for the original pipe image extracted by the industrial camera, the two-dimensional Otsu threshold segmentation method based on the improved firefly algorithm is used to automatically extract the pipe piece region of interest, then the two-dimensional inter-class variance of the image is taken as the objective function, and the particle swarm algorithm (PSO), grey wolf algorithm (GWO) and classical firefly algorithm (FA) are used to search for the maximum value of the two-dimensional inter-class variance, and the optimal individual when the maximum value of the two-dimensional inter-class variance is obtained is taken as the optimal two-dimensional threshold for segmentation, and the pipe piece region is segmented according to the optimal two-dimensional threshold, and compared with the scheme of the application, the comparison methods are named as: classical 2-Otsu (classical two-dimensional Otsu threshold segmentation method), PSO-2-Otsu (two-dimensional Otsu threshold segmentation method optimized by PSO), GWO-2-Otsu (two-dimensional Otsu threshold segmentation method optimized by GWO), FA-2-Otsu (two-dimensional Otsu threshold segmentation method optimized by FA), and the final results are shown in (a)-(f) of Figure 2 As can be seen, the scheme of the application extracts a more complete region in the extraction of the pipe piece region, and the noise is better removed, and has better performance.

[0060] The superiority of the method of the application is further investigated using the two numerical indicators of image Shannon entropy and region contrast.

[0061] The size of the image Shannon entropy reflects the degree of image segmentation, and the greater the entropy value, the better the segmentation effect, and the specific formula is as follows:

[0062] H = -P0log 10 P0-P1 log 10 P1

[0063] Wherein P0 and P1 represent the probabilities of 0 and 1 appearing in the binary image after segmentation respectively. The specific calculation results are shown in Table 1.

[0064] Table 1 Comparison of image Shannon entropy values

[0065]

[0066] The region contrast also reflects the image segmentation quality, and the formula is as follows:

[0067]

[0068] Wherein f0 and f1 represent the number of background pixel points and the number of target pixels respectively, and the specific calculation results of the region contrast are shown in Table 2.

[0069] Table 2 Region contrast results

[0070]

[0071] As shown in Table 1 and Table 2, the method has the optimal image segmentation and noise weakening performance in the two numerical indicators of image Shannon entropy and image contrast, and is superior to all the comparison algorithms. After the segmentation of the shield tail gap interest region is completed, the shield tail gap monitoring system based on machine vision extracts the calibration information in the interest region through image processing technology, obtains the shield tail gap according to the calibration information and a calculation formula, and finally displays the shield tail gap on a system graphical interface.

Claims

1. A two-dimensional Otsu thresholding method for scutocele gap images based on an improved firefly algorithm, characterized in that, include: Acquire images of the pipe segments; Randomly generated within the grayscale range [0, 255] of the pipe segment image. N A population consisting of only two-dimensional fireflies; The method for obtaining the optimal two-dimensional firefly position in a population through iteration includes: calculating the maximum brightness of each firefly in the population; and defining an objective function. f Two-dimensional inter-class variance values : in These represent the two-dimensional pixel probabilities of the foreground and background of the tube segment image, respectively. , This is a two-dimensional gray-level mean vector of the foreground and background of the tunnel segment image. Given the global two-dimensional gray-level mean vector of the pipe segment image, calculate the objective function. f The value of is used as the maximum brightness of the firefly; The fireflies in the population are ranked by their maximum brightness. For each firefly, it is grouped with several fireflies with lower brightness to form a local community. Attraction operations between fireflies were performed on local communities, and the relative brightness and attraction of each firefly to other fireflies in its local community were calculated. The firefly locations are updated based on relative brightness and attractiveness, and an attractiveness weight is introduced during the location update to strengthen the connection between local communities. Update the firefly's maximum brightness based on its new position. After multiple iterations, the optimal two-dimensional firefly position is output: in x j ( t +1) represents the position of firefly j at iteration number t+1. x i ( t Let ) represent the position of firefly i at iteration number t. x j ( t Let ) represent the position of firefly j at iteration number t. A ij ( r ij Let be the attraction between two fireflies i and j. α The step size factor is rand(), which represents the standard normal distribution of (0,1). x best ( t () represents the optimal individual position within the current local community. ω ( t ) represents the attraction weight of fireflies; in, f (x best (t) is the global optimal value in the t-th iteration. f (x best (t-1)) is the global optimal value in the (t-1)th iteration. f i (t-1) represents the firefly x in the (t-1)th iteration. i The value, f i (t-2) represents the firefly x in the (t-2)th iteration. i The value; The mean value of the optimal two-dimensional firefly position is taken as the best threshold. The best threshold is used to perform threshold segmentation on the grayscale image of the tube segment, thereby obtaining the segmentation result of the region of interest of the tube segment.

2. A two-dimensional Otsu thresholding system for scutocele gap images based on an improved firefly algorithm, characterized in that, include: An image acquisition module, configured to acquire images of the tube segment; The population initialization module is configured to randomly generate [a population] within the grayscale range [0, 255] of the tube segment image. N Two-dimensional fireflies A population of insects; The firefly attraction model is configured as follows: Calculate the maximum brightness of each firefly in the population: Define the objective function. f Two-dimensional inter-class variance values : in , These represent the two-dimensional pixel probabilities of the foreground and background of the tube segment image, respectively. This is a two-dimensional gray-level mean vector of the foreground and background of the tunnel segment image. Given the global two-dimensional gray-level mean vector of the pipe segment image, calculate the objective function. f The value of is used as the maximum brightness of the firefly; The fireflies in the population are ranked by their maximum brightness. For each firefly, it is grouped with several fireflies with lower brightness to form a local community. Attraction operations between fireflies were performed on local communities, and the relative brightness and attraction of each firefly to other fireflies in its local community were calculated. The firefly locations are updated based on relative brightness and attractiveness, and an attractiveness weight is introduced during the location update to strengthen the connection between local communities. Update the firefly's maximum brightness based on its new position. After multiple iterations, the optimal two-dimensional firefly position is output: in x j ( t +1) represents the position of firefly j at iteration number t+1. x i ( t Let ) represent the position of firefly i at iteration number t. x j ( t Let ) represent the position of firefly j at iteration number t. A ij ( r ij Let be the attraction between two fireflies i and j. α The step size factor is rand(), which represents the standard normal distribution of (0,1). x best ( t () represents the optimal individual position within the current local community. ω ( t ) represents the attraction weight of fireflies; in, f (x best (t) is the global optimal value in the t-th iteration. f (x best (t-1)) is the global optimal value in the (t-1)th iteration. f i (t-1) represents the firefly x in the (t-1)th iteration. i The value, f i (t-2) represents the firefly x in the (t-2)th iteration. i The value; The image segmentation module is configured to take the mean of the optimal two-dimensional firefly positions as the best threshold, and use the best threshold to perform threshold segmentation on the grayscale image of the tube segment, thereby obtaining the segmentation result of the region of interest of the tube segment.

3. An apparatus, characterized in that, include: processor; The memory includes one or more program modules; wherein the one or more program modules are stored in the memory and configured to be executed by the processor, and the one or more program modules include instructions for implementing the two-dimensional Otsu thresholding method for shield-tail gap images based on the improved firefly algorithm as described in claim 1.

4. A storage medium for storing non-transitory instructions, characterized in that, When the non-temporary instructions are executed by the processor, the two-dimensional Otsu threshold segmentation method for shield-tail gap images based on the improved firefly algorithm described in claim 1 can be implemented.

Images