A method for judging homogeneity of concrete based on 5G camera shooting

Abstract

The present application relates to a kind of based on 5G camera's concrete light darkness judgment homogeneity method, the method can be according to the inherent physical characteristics of concrete extraction, using algorithm mapping to computer data, obtain the physical law in image feature extraction, further improve the accuracy of image feature recognition technology;The method will also important liquidity and cohesiveness physical characteristics in concrete homogeneity Abstracted as numerical to judge concrete homogeneity, using lightness, darkness algorithm respectively on liquidity and cohesiveness characteristics are extracted specifically.The present application obtains concrete homogeneity state information by real-time detection to concrete mixture image, realizes real-time monitoring concrete homogeneity state.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of concrete mixing detection, and relates to a method for judging homogeneity of concrete lightness and darkness based on 5G camera shooting. BACKGROUND

[0002] With the rapid development of China's infrastructure industry, China has become a major producer of concrete and foundation stabilizing materials. The process of batching and mixing has an important influence on the quality of concrete. Therefore, developing and designing a concrete detection and measurement method suitable for complex environments can more effectively obtain high-quality concrete, which plays an important role in improving the efficiency and quality of concrete mixing, saving labor costs, stabilizing the quality of concrete mixture, and ensuring normal and smooth construction of projects.

[0003] Concrete homogeneity refers to the relative error of the volume density of the mortar in the concrete being less than 0.8%, and the relative error of the weight of the coarse aggregate in the unit volume of concrete being less than 5%. Uniformly mixed concrete mixture is a prerequisite for ensuring that the performance of hardened concrete meets the requirements. Currently, there is no method for real-time detection of concrete homogeneity during the mixing process. Traditional manual detection methods lack convenience in today's large demand and supply, and cannot guarantee the reliability of the detection results. Modern concrete homogeneity measurement methods mainly involve observing the flow state of the mortar after mixing by allowing the concrete to freely fall on a horizontal surface. This method is not real-time because it is performed after mixing. In order to achieve real-time homogeneity monitoring of concrete mixing, effective data information can be obtained by combining AI intelligent vision processing based on the physical properties of concrete homogeneity.

[0004] Artificial intelligence (AI) is a new technology science that studies, develops, simulates, extends, and expands human intelligence. Camera, as the eyes of the machine, has the characteristics of low price and simple configuration. With the continuous improvement of image processing technology and the development of information transmission technology, information in multimedia data such as video and image is being continuously mined to provide a basis for further decision-making.

[0005] 5G is a new network architecture with high speed, ubiquitous network, low power consumption, and low latency, which can provide higher peak rate, better mobile performance, millisecond-level latency, and ultra-high density connection. 5G technology will bring great changes to the vertical business field of intelligent construction of concrete mixing. It is imperative to develop high-end intelligent concrete mixing based on 5G technology.

[0006] The application provides a concrete light-darkness homogeneity AI measurement method and system based on 5G camera shooting, which obtains the image of the concrete mixer shot by the camera through 5G communication technology, and processes the video stream frame by frame, first pre-processes the key area, then extracts the concrete mixture fluidity state value through the lightness algorithm and the darkness algorithm extracts the concrete mixture cohesiveness state value, combines AI intelligent fusion to detect the concrete mixture image in real time, so as to obtain the concrete homogeneity state information, and realize real-time monitoring of the concrete homogeneity state. SUMMARY

[0007] With the gradual upgrading of network bandwidth, camera video acquisition is rapidly developing in the field of image recognition. Existing cases prove that camera-based image recognition technology has great advantages in reducing labor costs, reducing on-site risks, enhancing enterprise technological competitiveness and the like.

[0008] The fluidity and cohesiveness in the concrete workability are used to judge whether the concrete is mature in the mixing process and to calculate the slump value. Among the four basic components of ordinary concrete, water, cement, sand (fine aggregate) and gravel (coarse aggregate). According to the paste film thickness theory, concrete can be regarded as a two-phase material composed of mortar and coarse aggregate. The rheological properties of mortar and film thickness determine the workability of concrete. Only when the mortar completely fills the gap between the coarse aggregate and forms a good wrapping layer around the coarse aggregate particles will the concrete produce a certain fluidity. If the homogeneity of the concrete mixture does not meet the requirements, it will inevitably lead to uneven distribution of the components of the concrete, resulting in segregation phenomenon, uneven shrinkage during hardening, thus increasing the cracking probability and reducing the engineering quality. In order to accurately and timely identify and intelligently judge the mixture in a complex environment, it is necessary to intelligently monitor the surrounding environment and capture the mixing picture, extract the typical features, perceive the effective information of the mixture in the mixing tank, and make corresponding judgments according to different states to obtain real-time mixture state information. Therefore, the concrete homogeneity detection technology in a complex environment is one of the important environments to ensure the work performance of concrete.

[0009] In order to solve the existing problems, the application provides a kind of concrete light-darkness homogeneity AI measurement method and system based on 5G camera, which comprises the following steps: using high-definition camera to shoot the video of concrete mixture being stirred, transmitting in real time through 5G network, extracting key area in image frame sequence in monitoring area on host computer, starting preliminary processing of image, and image preprocessing includes acquisition, grayscale, image enhancement and binarization processing and other links.Then firstly, the flowability characteristics of concrete mixing are extracted, firstly, the luminosity change is carried out, and the flowability of concrete mixture is highlighted;Next, through the darkness change of image, the cohesiveness characteristics of image are extracted, the condition number that satisfies two main physical characteristics is counted, and finally the uniformity of concrete is determined according to the nature of homogeneity.

[0010] The technical scheme of the application relates to two aspects:

[0011] On the one hand, according to the inherent physical characteristics of concrete, the algorithm is mapped to computer data, the physical law in image feature extraction is obtained, and the accuracy of image feature recognition technology is further improved;

[0012] On the other hand, a method for abstracting the flowability and cohesiveness physical characteristics important in concrete homogeneity into numerical values to judge the homogeneity of concrete is proposed, and the flowability and cohesiveness characteristics are extracted by using luminosity and darkness algorithms respectively.

[0013] The flowability of concrete refers to the performance that concrete mixture flows under the action of its own gravity or machine vibration, and can uniformly flow into the formwork, which is crucial to the dilution degree of concrete mixing.

[0014] According to the rheological properties of mortar, water and cement form cement slurry, and mortar affects the flowability of self-compacting concrete by affecting the stress and movement conditions of coarse aggregate particles. Concrete with sufficient mixing time shows good flowability in the mixing tank, and concrete without mixing shows uneven particle shape.

[0015] A method for extracting flowability characteristic numerical value by luminosity comprises the following steps:

[0016] Step one, image light channel generation; the value of the brightest color in the RGB three channels of true color image is stored as luminosity in the light channel generation image, as shown in formula (1)

[0017]

[0018] In the formula, Jc represents a color channel, and Ω (x) is a 3*3 square area centered at x, and x represents a color value.

[0019] Step two, guiding filter filters the image; the basic concept is: given the input image p, the guide image I, the output image is obtained, the input image p and the guide image I can be the same image. The image after filtering can be expressed by the weighting formula as (2)

[0020]

[0021] In the formula, i is the index of the current filtering point, j is the index of all pixel points covered by the filter template, the filter (convolution kernel) W ij is a function of the guide image I, which is independent of the input image p.

[0022] The image after guiding filtering can clearly display the flowability of the concrete in the computer. If the flowability of the concrete is good, the image shows that the paste is uniformly wrapped, and there is no obvious stone exposed, and the whole image is a good gray image. If the flowability of the concrete is poor, the image shows that the stone is not wrapped by the paste.

[0023] Step three, black and white the filtered image; after obtaining the filtered image representing the flowability of the concrete, further black and white the image can obtain the image containing only the stone not wrapped by the concrete paste. The more white points in the image, the worse the flowability of the concrete. First, the true color image is converted into a gray scale image. As an optimization, the weighted average method is used for gray scale processing. Studies have shown that people have different sensitivities to different colors, with the highest sensitivity to green, followed by red, and the lowest sensitivity to blue, as shown in formula (3)

[0024] E'y = 0.299R + 0.587G + 0.114B (3)

[0025] In the formula, E'y is the gray scale image, and R, G and B represent the red, green and blue color values of the three channels respectively.

[0026] After the image is grayed, the threshold value is obtained by using the maximum inter-class variance method, and the threshold value can be used to convert the gray scale image into the best black and white image, wherein the white point represents the stone not wrapped by the paste.

[0027] Step four, statistics image feature value; after obtaining the black and white, the proportion of white points in the image is counted, and the flowability characteristic value is obtained. The larger the flowability value, the fewer the white points, that is, the fewer the stones not wrapped by the paste, and the better the flowability.

[0028] The above photometric extraction flowability characteristic method can effectively extract the flowability characteristics as much as possible, and the flowability of the concrete mixing can be expressed through the image.

[0029] The cohesiveness of concrete refers to the performance of the concrete mixture in the construction process, i.e., the performance of keeping the combination between the constituent materials without delamination and segregation.

[0030] Under the observation of the image taken by the high-definition camera, the cement bonding wraps the coarse and fine aggregates to form a whole, fills the gaps between the aggregates, and improves the compactness. However, the unripe concrete has stones exposed outside, which presents obvious local abnormal color on the image.

[0031] The method for extracting the cohesiveness feature value from darkness includes the following steps.

[0032] Step 1: generating the dark channel of the image; the darkest value of the color in the RGB three channels of the true color image is stored as the luminosity to generate the dark channel of the image, as shown in formula (4).

[0033]

[0034] In the formula, Jc represents a color channel, and Ω(x) is a square region centered at x, and x represents a color value.

[0035] Step 2: filtering the image by using the guided filter; the step is consistent with the step 1 for extracting the liquidity feature value from the luminosity.

[0036] Step 3: calculating the reflection component of the image. First, the illumination component estimation is performed on the whole image of each color channel, then the illumination component is subtracted, and finally the gray scale is stretched, so that the target image is obtained. The target image contains the cohesiveness feature information of the concrete. In the image, the dark spots represent the large particles of the concrete that have not been stirred uniformly. The more the dark spots, the worse the cohesiveness of the concrete, and the less uniform the stirring.

[0037] Step 4: counting the feature value of the image. After obtaining the feature image containing the cohesiveness information of the concrete, the physical feature information of the cohesiveness of the concrete is further quantified, and the proportion of the dark spots in the whole image is counted. The greater the cohesiveness feature value, the greater the particle size of the concrete that has not been stirred uniformly, and the worse the cohesiveness of the concrete.

[0038] As a preferred embodiment, the luminosity processing is performed on the extracted image, and the liquidity feature in the homogeneity is abstracted into a specific value.

[0039] As a preferred embodiment, the darkness processing is performed on the extracted image, and the cohesiveness feature in the homogeneity is abstracted into a specific value.

[0040] As a preferred embodiment, the liquidity feature value and the cohesiveness feature value are obtained through repeated experiments, and the best and most optimal parameter range is obtained. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to have a more intuitive understanding of the embodiments of the present application, the drawings required by the embodiments will be briefly introduced as follows. In the drawings:

[0042] Figure 1 Flow chart for extracting the lightness feature value of the fluidity of the present application;

[0043] Figure 2 Flow chart for extracting the darkness feature value of the cohesiveness of the present application;

[0044] Figure 3 Schematic diagram of the lightness and darkness guided filter model of the present application;

[0045] Figure 4 Flow chart for imaging the fluidity algorithm;

[0046] Figure 5 Flow chart for judging the homogeneity of the concrete according to the feature fusion of the present application; DETAILED DESCRIPTION

[0047] The technical solutions will be further described in detail below in combination with the drawings in the examples of the present application.

[0048] Figure 1 Flow chart for extracting the lightness feature value of the fluidity of the present application.

[0049] In combination with the drawings Figure 1 , the following steps are included:

[0050] Step one, image light channel generation; the value of the color with the brightest light in the RGB three channels of the true color image is stored as the lightness into the light channel generation diagram, as shown in formula (1)

[0051]

[0052] In the formula, Jc represents a certain color channel, and Ω(x) is a 3*3 square area centered at x, and x represents the color value.

[0053] Step two, guided filter filtering of the image; the basic concept is: given an input image p and a guide image I, an output image is obtained, and the input image p and the guide image I can be the same image. The image after filtering can be represented by the weighting formula (2)

[0054]

[0055] In the formula, i is the index of the current filtering point, j is the index of all pixel points covered by the filter template, and the filter (convolution kernel) W ij is a function of the guide image I, which is independent of the input image p.

[0056] The image after the guided filtering can be displayed clearly in the computer to show the flowability of the concrete. If the flowability of the concrete is good, the image shows that the paste uniformly wraps the stones, and no stone is exposed, and the whole image is a good gray image. If the flowability of the concrete is poor, the image shows that there are obvious white points, which indicates that the stones are not wrapped by the paste.

[0057] Step three, black and white of the filtered image; after obtaining the filtered image showing the flowability of the concrete, the image is further black and white, so that only the image of the stones not wrapped by the concrete paste is obtained. The more white points in the image, the poorer the flowability of the concrete. First, the true color image is converted into a gray scale image. As preferred, the weighted average method is used for the gray scale processing. Studies have shown that people have different sensitivities to different colors, the highest sensitivity to green, the second highest sensitivity to red, and the lowest sensitivity to blue, as shown in formula (3).

[0058] E'y = 0.299R + 0.587G + 0.114B (3)

[0059] In the formula, E'y is the gray scale image, and R, G and B represent the color values of the red, green and blue channels respectively.

[0060] After the image is converted into a gray scale image, the threshold value is obtained by using the maximum inter-class variance method, and the threshold value can be used to convert the gray scale image into an optimal black and white image, in which the white points represent the stones not wrapped by the paste.

[0061] Step four, statistics of image feature values; after the black and white is obtained, the proportion of white points in the image is counted, so that the flowability feature value is obtained. The larger the flowability value, the fewer the white points, that is, the fewer the stones not wrapped by the paste, and the better the flowability.

[0062] Figure 2 The flow chart for extracting the adhesion feature value by the darkness of the present application.

[0063] The accompanying drawings Figure 2 comprise the following steps:

[0064] Step one, image dark channel generation; the value of the darkest color in the RGB three channels of the true color image is stored as the luminosity to generate a light channel image, as shown in formula (4).

[0065]

[0066] In the formula, Jc represents a color channel, and Ω(x) is a square region centered at x, and x represents a color value.

[0067] Step two, guided filter filtering image; the same as step one of the luminosity extraction flowability feature value.

[0068] Step three, calculate the reflection component of the image. First, estimate the illumination component of the full image for each color channel, then subtract the illumination component, and finally stretch the gray scale to obtain the target image. The target image contains the cohesive feature information of the concrete, and the dark spots in the image represent large concrete particles that have not been mixed uniformly. The more dark spots, the worse the cohesive properties of the concrete and the less uniform the mixing;

[0069] Step four, count the image feature value. After obtaining the feature image containing the cohesive information of the concrete, further number the physical feature information of the cohesive of the concrete, and count the proportion of dark spots in the image as a whole. The larger the cohesive feature value, the larger the particle size that has not been mixed uniformly, and the worse the cohesive properties of the concrete.

[0070] Figure 3 The figure is a luminosity guided filtering model schematic diagram of the present application, (a) is a luminosity guided filtering model schematic diagram, the concrete mixture undergoes complex chemical reaction changes during the mixing process, and the fluidity changes continuously with the mixing process. After the luminosity guided filtering of the algorithm of the present application, the image preliminarily presents the concrete fluidity representation, mainly highlighting the number and position of stones that have not been wrapped by concrete mortar, and directly showing the image white point of abnormal color; (b) is a dark guided filtering model schematic diagram, as one of the three physical properties of concrete mixture, cohesive feature is very important for judging the quality of concrete. The original picture is darkened to obtain the cohesive feature of the concrete. The more black spots in the image, the larger the particle group that has not been fused and coagulated, and the worse the cohesive properties.

[0071] Figure 4 It is a flowability algorithm imaging diagram. After further processing by the algorithm, the image is a binary image, and the physical characteristics of the concrete fluidity are further highlighted. The white spots in the image are the images of stones that have not been wrapped by mortar. The more white spots, the worse the fluidity of the image. The proportion of white spots in the image is calculated with the size of the image to obtain the judgment of the fluidity feature.

[0072] Figure 5 It is a method for judging the homogeneity of concrete light and darkness based on 5G camera of the present application, comprising the following steps:

[0073] S1: import the picture to be processed by 5G signal transmission;

[0074] S2: after measuring the concrete fluidity feature parameter by the method for extracting the fluidity feature value of the present application, exclude the interference of the blade;

[0075] S3: after excluding the above interference, continue to measure the cohesive feature parameter of the concrete by the method for extracting the cohesive feature value of the present application to judge;

[0076] S4 extracts two important characteristic parameters of luminosity and darkness related to the homogeneity of concrete respectively, judges whether it is in the luminosity characteristic range and the darkness characteristic range, and obtains the homogeneity state of the concrete by fusing the characteristic values.

[0077] The protection scope of the present application is not limited to the above-mentioned embodiments, and any technical solution falling within the concept of the present application falls within the protection scope of the present application. It should be noted that, for ordinary skilled persons in the technical field, some improvements and refinements without departing from the principle of the present application should also be considered as falling within the protection scope of the present application.

Claims

1. A method for extracting luminosity flowability feature value, comprising the following steps: Step one, image light channel generation; the value of the brightest color in the RGB three channels of the true color image is stored as luminosity in the light channel generation image, as shown in formula (1): （1） where Jc represents a certain color channel, and is a 3*3 square region centered at and z represents the color value. Step two, guided filter filtering image; the basic concept is: given input image p, guide image I, get output image, input image p and guide image I are the same image; the image after filtering can be represented by the weighted formula (2): （2） where i is the index of the current filtering point, j is the index of all the pixel points covered by the filter template, the filter is a function of the guiding image I, independent of the input image p: The image after guided filtering can clearly display the flowability state of concrete in the computer, if the flowability of concrete is good, it shows that the paste is uniformly wrapped, there is no obvious stone exposed, and the whole image shows a good gray image; if the flowability of concrete is poor, it shows that the stone has not been wrapped by the paste; Step three, filtering image black and white; after obtaining the filtering image representing the flowability of concrete, further black and white image can be obtained, which only contains the stone image not wrapped by the concrete paste, the more white points in the image, the worse the flowability of concrete; first, the true color image is converted into a gray scale image, and the weighted average value method is used for gray scale processing: as shown in formula (3): E'y = 0.299R + 0.587G + 0.114B (3) In the formula, E'y is the gray scale image, R, G and B represent the color values of red, green and blue three channels respectively; After image gray scale, the threshold value is obtained by using the maximum interclass variance method, and the threshold value can be used to convert the gray scale image into the best black and white image, in which the white point represents the stone not wrapped by the paste; Step four, statistical image feature value; after obtaining the best black and white image, the proportion of white points in the image is counted, and the flowability feature value is obtained, the larger the flowability value, the fewer the white points, that is, the fewer the stones not wrapped by the paste, and the better the flowability.

2. A method for extracting darkness cohesion feature value, comprising: Step one, image dark channel generation; the value of the darkest color in the RGB three channels of the true color image is stored as luminosity in the light channel generation image, as shown in formula (4): （4） where Jc represents a certain color channel, and is a square region centered at with z representing the color value. Step two, guided filter filtering image; the basic concept is: given input image p, guide image I, get output image, input image p and guide image I are the same image; the image after filtering can be represented by the weighted formula (2): （2） In the formula, i is the index of the current filtering point, j is the index of all pixel points covered by the filter template, and the filter Wij is a function of the guide image I, which is independent of the input image p: The image after guided filtering can clearly display the cohesion state of concrete in the computer, if the cohesion of concrete is good, it shows that the paste is uniformly wrapped, there is no obvious stone exposed, and the whole image shows a good gray image; if the cohesion of concrete is poor, it shows that the stone has not been wrapped by the paste; Step three, calculate the reflection component of the image; first, estimate the illumination component of the full image for each color channel, then subtract the illumination component, and finally stretch the gray scale to obtain the target image; the target image contains concrete cohesion feature information, and the dark spots in the figure represent large concrete particles that have not been mixed evenly, the more dark spots, the worse the concrete cohesion, and the less uniform the mixing; Step four, count the image feature value; after obtaining the feature image containing the concrete cohesion information, further number the concrete cohesion physical feature information, and count the proportion of dark spots in the overall image. The larger the cohesion feature value, the larger the unblended particle size, and the poorer the concrete cohesion.

3. A method for judging the homogeneity of concrete based on 5G camera shooting, comprising the following steps: S1: import the picture to be processed through 5G signal transmission picture; S2: after measuring the concrete fluidity characteristic parameter by the method for extracting the fluidity characteristic value according to claim 1, exclude the blade interference; S3: after excluding the above interference, continue to measure the concrete cohesion characteristic parameter by the method for extracting the cohesion characteristic value according to claim 2 to judge; S4: after extracting the two important characteristic parameters of lightness and darkness related to the homogeneity of concrete, judge whether it is in the lightness characteristic range and the darkness characteristic range, and obtain the homogeneity state of concrete by fusing the characteristic values.

Images