A machine vision-based concrete uniformity detection method and system

Abstract

The application belongs to the technical field of image processing, and particularly relates to a concrete distribution uniformity detection method and system based on machine vision, which comprises the following steps: collecting a concrete distribution image to be detected and performing smoothing processing; for any pixel point in the processed image: determining a local gray fluctuation energy according to the spatial distance and gray difference between the pixel point and the pixel points in its neighborhood; constructing a flatness index combining the gray mean and variance of all the neighborhood pixel points of the pixel point; determining the distribution uniformity according to the mean and standard deviation of the flatness indexes of all the neighborhood pixel points of the pixel point; and marking the non-uniform points of the distribution based on the experience threshold determined based on the distribution of qualified samples. The application can effectively distinguish normal mechanical stripes from abnormal accumulation, and improve the accuracy and anti-interference ability of the concrete distribution uniformity detection.

Description

Technical Field

[0001] This invention relates to the field of image processing technology. More specifically, this invention relates to a machine vision-based method and system for detecting the uniformity of concrete placement. Background Technology

[0002] In the production process of precast concrete components, the uniformity of material distribution is the key to quality control. Due to the non-Newtonian fluid properties of concrete slurry, local clumps or material shortages are easily formed during the pouring process. Under machine vision imaging, these non-uniform areas appear as significant surface undulations and produce strong light and shadow contrasts under lighting conditions. That is, the highlight surface and the shadow surface appear as local drastic gray-scale jumps in the image. Therefore, it is necessary to capture the gray-scale change characteristics in the image to determine whether the material distribution is uniform.

[0003] Local Binary Pattern (LBP) encodes local microscopic grayscale relationships into a binary pattern by comparing the grayscale values ​​of the center pixel with those of its neighboring pixels, thereby describing texture features. Because LBP is highly sensitive to grayscale fluctuations in local areas, it can effectively capture minute edge and undulation information in images, and is therefore often used in surface uniformity detection.

[0004] However, concrete placing booms leave behind regular, high-frequency transverse roughening stripes during operation. These normal mechanical stripes also appear as continuous peaks and troughs on a microscopic scale, forming dense local gray-scale jumps. For local binary models, the gray-scale differences produced by normal stripes and those produced by abnormally accumulated clumps are extremely similar in their local neighborhoods. This causes the local binary model to be unable to distinguish between directional normal stripe undulations and irregular abnormal accumulation undulations, misjudging normal stripes in the entire image as uneven areas and seriously interfering with the accuracy of the concrete placement uniformity detection results. Summary of the Invention

[0005] To address the technical problem that the aforementioned local binary model cannot distinguish between normal roughening stripes on the concrete surface and abnormal fabric distribution, resulting in inaccurate fabric distribution uniformity detection results, this invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a machine vision-based method for detecting the uniformity of concrete placement, comprising: acquiring an image of concrete placement to be detected and performing smoothing processing to obtain the grayscale value and position of each pixel in the processed image; for any pixel in the processed image: calculating a weighted average of the absolute values ​​of the grayscale differences between the pixel and its neighboring pixels based on the spatial distance between the pixel and its neighboring pixels to determine the local grayscale fluctuation energy; calculating the mean and variance of the grayscale values ​​of all neighboring pixels of the pixel; determining a flatness index based on the local grayscale fluctuation energy of the pixel and the mean and variance; determining the uniformity of placement based on the mean and standard deviation of the flatness index of all neighboring pixels of the pixel; obtaining an empirical threshold determined based on the distribution of qualified samples; identifying pixels in the image of concrete placement to be detected whose uniformity of placement is less than the empirical threshold and marking them as non-uniform placement points.

[0007] This invention constructs local grayscale fluctuation energy by introducing spatial distance weighting, assigning higher weights to neighboring pixels to effectively capture microscopic texture abrupt changes caused by accumulation or material shortage. Furthermore, it combines the mean and variance of the neighborhood grayscale to construct a flatness index, utilizing the characteristic that normal mechanical roughening stripes, although having a large variance, have a stable mean, to construct suppression weights to distinguish and suppress normal texture interference, highlighting real fabric anomalies. Simultaneously, it determines fabric uniformity by calculating the mean and normalized standard deviation of the flatness index, effectively suppressing high-frequency discrete noise by utilizing the low-value clustering characteristics of real defects. Finally, based on the distribution of qualified samples, it determines an empirical threshold, achieving accurate detection of concrete fabric uniformity under different environments, solving the problem that traditional methods struggle to distinguish between normal textures and abnormal accumulation.

[0008] Preferably, the step of acquiring and smoothing the concrete placement image to be detected includes: converting the concrete placement image to grayscale to obtain a grayscale image; and performing Gaussian filtering on the grayscale image to obtain a processed image.

[0009] Preferably, the local grayscale fluctuation energy satisfies the expression: In the formula, For the first Local grayscale fluctuation energy of each pixel; For the first The grayscale value of each pixel; For the first The first pixel The grayscale value of each neighboring pixel; For the first The pixel and its first Euclidean distance between neighboring pixels; This is the preset standard deviation parameter for spatial distance; These are the index values ​​and total number of neighboring pixels; To take the absolute value; It is a natural exponential function.

[0010] This invention utilizes the exponential decay characteristic to assign higher weights to neighboring pixels, enabling the calculation to focus on the micro-texture abrupt changes of the central pixel. At the same time, by optimizing the standard deviation parameter, it avoids excessive boundary truncation effects, effectively suppresses background interference caused by distant aggregate edges, and improves the ability to resolve fine aggregate edges and micro-textures within a limited neighborhood window.

[0011] Preferably, the flatness index satisfies the expression: In the formula, For the first Flatness index of each pixel; For the first Local grayscale fluctuation energy of each pixel; For the first The mean and variance of gray levels of the neighboring pixels of a pixel; These are preset hyperparameters; The average grayscale value of all pixels; To take the absolute value; It is a natural exponential function.

[0012] This invention utilizes the characteristic of large grayscale variance in normal napped stripes to automatically reduce their response value through a mathematical model. At the same time, it combines global mean deviation to accurately screen out areas with small variance but abnormal brightness, thereby effectively distinguishing normal mechanical stripes with high-frequency oscillation characteristics from real fabric defects that manifest as local smoothness anomalies, thus improving the accuracy of feature extraction.

[0013] Preferably, determining the fabric uniformity includes: normalizing the standard deviation of the flatness index of all neighboring pixels of each pixel to obtain a normalized standard deviation; the fabric uniformity is half the sum of the mean of the flatness index of all neighboring pixels of the pixel and the normalized standard deviation.

[0014] This invention utilizes the spatial clustering characteristic of real defects and introduces normalized standard deviation as a weight to calculate fabric uniformity. By amplifying the numerical difference between locally clustered defects and high-frequency discrete noise, this method effectively filters out artifact interference caused by light reflection or isolated aggregate edges, ensuring that the detection results accurately reflect the macroscopic fabric quality and enhancing the algorithm's noise resistance.

[0015] Preferably, obtaining the empirical threshold determined based on the distribution of qualified samples includes: acquiring multiple manually inspected and qualified concrete placement images as qualified samples; calculating the uniformity of placement of all pixels in all qualified samples and arranging them in ascending order, and selecting the value at the 5th position after arrangement as the empirical threshold.

[0016] This invention determines an empirical threshold by statistically analyzing the uniformity distribution characteristics of qualified samples, establishing a data-driven dynamic judgment standard that effectively reduces false detections and missed detections caused by environmental non-steady-state factors.

[0017] Preferably, the Gaussian filtering process uses a filter kernel size of [size missing]. .

[0018] Preferably, the method further includes: obtaining data centered on each pixel. All pixels within the neighborhood window are considered as the neighboring pixels of that pixel; where, It is an odd number greater than or equal to 3.

[0019] Preferably, the method further includes: calculating the ratio of the number of uneven points to the total number of pixels in the concrete placement image to be detected; if the ratio exceeds 3%, the current placement is determined to be uneven.

[0020] Secondly, the present invention provides a machine vision-based concrete uniformity detection system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned machine vision-based concrete uniformity detection method is implemented.

[0021] By adopting the above technical solution, a computer program for detecting the uniformity of concrete placement based on machine vision is generated and stored in a memory so that it can be loaded and executed by a processor. A terminal device can then be made based on the memory and processor for convenient use.

[0022] The beneficial effects of this invention are as follows:

[0023] This invention constructs a flatness index and utilizes a gray-scale variance suppression mechanism to effectively distinguish between normal regular roughening stripes and abnormal accumulation areas on the concrete surface, solving the technical problem that local binary modes easily misjudge normal stripes as uneven areas. It introduces the standard deviation of the local flatness index as a weight to calculate the uniformity of material distribution, effectively eliminating high-frequency discrete artifacts caused by light reflection or coarse aggregate edges by utilizing the spatial aggregation characteristics of real defects, ensuring that the detection results mainly reflect the macroscopic quality of material distribution. Furthermore, it employs a dynamic empirical threshold strategy based on multi-sample distribution statistics to replace the traditional fixed threshold judgment, enabling the system to adapt to fluctuations in different batches of concrete materials and lighting environments, reducing the rate of missed detections and false alarms caused by environmental changes. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating a machine vision-based method for detecting the uniformity of concrete placement in this invention.

[0025] Figure 2 It is a schematic grayscale image showing the concrete fabric to be inspected;

[0026] Figure 3 This is a schematic diagram illustrating the brightness distribution of the concrete distribution to be tested. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0029] This invention discloses a machine vision-based method for detecting the uniformity of concrete placement, referring to... Figure 1 This includes steps S1-S5:

[0030] S1. Acquire the image of the concrete placement to be detected and perform smoothing processing to obtain the grayscale value and position of each pixel in the processed image.

[0031] It should be noted that, in order to accurately monitor the overall distribution of the fabric area, images were acquired by taking pictures from a vertical top-down position to ensure that the obtained concrete surface texture has isotropic scale characteristics in spatial distribution. Considering that the inherent micro-roughness of the concrete slurry surface, such as sand particles and micro-bubbles, will form high-frequency random noise and affect the accuracy of subsequent feature analysis, filtering is required to improve image quality, suppressing micro-particle noise while fully preserving the macro-texture features of the fabric.

[0032] Specifically, on the precast concrete component production line, an industrial camera is mounted behind the concrete placing machine to take an overhead picture of the poured concrete as the concrete placement image to be inspected, and then the image is processed into grayscale to obtain a grayscale image.

[0033] For example, Figure 2 The image shows a grayscale image of the concrete fabric to be inspected, which includes normal lateral textured patterns and areas of uneven fabric distribution.

[0034] Furthermore, a Gaussian filter is used to smooth the grayscale image, obtaining the processed image and the grayscale value of each pixel in the image. Simultaneously, the position of each pixel also needs to be obtained. Specifically, the pixel at the bottom left corner of the image is taken as the origin, and the horizontal direction to the right from the origin is defined as... The positive direction of the axis is defined by taking the vertically upward direction from the origin as... The positive direction of the axis is used to construct a Cartesian coordinate system; the position of each pixel in the image is obtained in the Cartesian coordinate system, including... Coordinates in the axial direction and Coordinates along the axis.

[0035] In this embodiment of the invention, the Gaussian filter kernel is selected as... The personnel responsible for implementation may be adjusted according to the actual situation.

[0036] At this point, the grayscale value and position of each pixel in the processed image have been obtained.

[0037] S2. For any pixel in the image: Based on the spatial distance between the pixel and its neighboring pixels, perform a weighted average of the absolute values ​​of the grayscale differences between the pixel and its neighboring pixels to determine the local grayscale fluctuation energy.

[0038] It should be noted that the non-uniformity of concrete placement is mainly reflected in the disruption of grayscale continuity in local areas. For example, the shaded side of the pile will form a shadow, causing a sharp darkening of grayscale, while material shortages leading to exposed rebar will produce abrupt texture changes. However, as a composite material, the exposed edges of coarse aggregates on the surface of concrete will also produce significant grayscale differences. If only the grayscale difference is statistically analyzed, it will be difficult to distinguish between normal aggregate edges and abnormal pile edges. Therefore, introducing spatial distance factors for weighted constraints, by assigning higher weights to neighboring pixels, can more sensitively capture drastic changes in texture within a local micro-range, thereby assessing the local grayscale fluctuation energy of each pixel, providing a data foundation for subsequent removal of aggregate interference by combining regional statistical features.

[0039] Specifically, the neighboring pixels of each pixel are obtained as follows: Define a centered point on each pixel. The neighborhood window takes all pixels within the neighborhood window as the neighboring pixels of the given pixel.

[0040] in, The local analysis range of a pixel is defined to capture grayscale fluctuations within a local area. The value cannot be too small to avoid failing to capture the fluctuations. Furthermore, the analysis process involves acquiring neighboring pixels within a neighborhood window centered on the pixel. The value is an odd number greater than or equal to 3; this invention will The value is set to 5, and the implementers can adjust it as needed. It should be added that for pixels in the edge area where a complete neighborhood window cannot be obtained, the pixel's grayscale and position information are used for completion processing.

[0041] For any pixel, the local gray-level fluctuation energy of that pixel is determined based on its spatial distance from and gray-level differences with neighboring pixels; the local gray-level fluctuation energy satisfies the expression:

[0042]

[0043] In the formula, For the first Local grayscale fluctuation energy of each pixel; For the first The grayscale value of each pixel; For the first The first pixel The grayscale value of each neighboring pixel; For the first The pixel and its first Euclidean distance between neighboring pixels; This is the preset standard deviation parameter for spatial distance. Take 1.5; These are the index values ​​and total number of neighboring pixels; To take the absolute value; It is a natural exponential function.

[0044] in, Reflecting the The pixel and its first The grayscale difference between neighboring pixels; Reflecting the The first pixel The weight values ​​of each neighboring pixel decrease exponentially with increasing Euclidean distance, reflecting the constraint of spatial location on the contribution of grayscale changes. That is, the closer the neighboring pixel is to the center pixel, the greater its contribution to the local energy due to grayscale differences. The local grayscale fluctuation energy is obtained by multiplying the grayscale difference amplitude by the corresponding weight value and calculating the weighted average difference of all pixels in the neighborhood window. The larger the energy value, the more significant the grayscale discontinuity of the pixel in the micro-neighborhood. This discontinuity is mainly concentrated in the nearest neighbor area, which can effectively highlight the texture abrupt features caused by accumulation or lack of material and suppress background interference caused by distant aggregate edges.

[0045] It should be added that the standard deviation parameter of spatial distance The value range is usually set between 0.8 and 2.5, based on the Gaussian distribution. Guidelines Values ​​between 1.5 and 1.7 generally yield good results; in this embodiment, 1.5 is selected to avoid [further issues]. Excessively large Gaussian weights result in an overly flat Gaussian weight distribution, thus introducing an excessively large boundary truncation effect, ensuring that... Within a limited neighborhood window, energy calculation can focus more on the micro-texture mutations of the central pixel, improving the ability to resolve the edges of fine aggregates.

[0046] At this point, the local grayscale fluctuation energy of each pixel has been obtained.

[0047] S3. Calculate the mean and variance of the gray values ​​of all neighboring pixels of the pixel; determine the flatness index based on the local gray-level fluctuation energy of the pixel and the mean and variance.

[0048] It should be noted that due to the high-frequency transverse fraying stripes generated during the concrete application process, the local grayscale fluctuation energy of the corresponding pixels in this area is also high, making it difficult to distinguish between uneven application areas and normal stripe areas, thus affecting detection accuracy. In reality, uneven application usually involves accumulation or pitting, manifesting as a deviation in the overall brightness of a local area, and this deviation has a certain degree of smoothness. The mechanical fraying stripes during the application process, on the other hand, are high-frequency oscillations. Therefore, a suppression weight can be constructed by combining the grayscale mean and variance of a local area. If a region has a large grayscale variance but a small change in mean, it should be judged as normal texture and suppressed. If the grayscale variance is small but the difference from the background is large, uneven application may exist.

[0049] Specifically, the mean and variance of gray levels of all neighboring pixels of any given pixel are calculated; based on the mean and variance of gray levels of the neighboring pixels of any given pixel, and the local gray level fluctuation energy of any given pixel, the flatness index of that pixel is determined; the flatness index satisfies the expression:

[0050]

[0051] In the formula, For the first Flatness index of each pixel; For the first Local grayscale fluctuation energy of each pixel; For the first The mean and variance of gray levels of the neighboring pixels of a pixel; These are preset hyperparameters; The average grayscale value of all pixels; To take the absolute value; It is a natural exponential function.

[0052] in, Reflecting the The suppression weight of each pixel is determined by the fact that the larger the gray-level variance of the neighboring pixels of a pixel, the more likely the pixel is to be located in the normal stripe area. This will suppress the local gray-level fluctuation energy of the pixel, thereby reducing the response value of the pixel and suppressing normal texture noise. This reflects the degree to which the local brightness of a pixel deviates from the global background. A larger value indicates that the pixel is more likely to belong to an area of ​​uneven fabric distribution. In summary, if... The larger the value and The larger the value, the more likely the pixel is to belong to a region of uneven fabric texture, and the more likely it is to have a significant brightness anomaly. This means the pixel's flatness index is lower. To preset hyperparameters, We set it to 0.1 to avoid the denominator being 0.

[0053] At this point, the flatness index of each pixel has been obtained.

[0054] S4. Determine the fabric uniformity based on the mean and standard deviation of the flatness index of all neighboring pixels of the pixel.

[0055] It should be noted that real uneven areas of concrete distribution, such as local bulges caused by accumulation or pits and shadows caused by material shortage, are usually characterized by low clusters of flatness index values ​​with small fluctuations in the local area. In contrast, artifacts caused by light reflection or coarse aggregate edges, although having a low flatness index, usually exhibit a high-frequency discrete distribution, resulting in a large local variance. Therefore, in order to further enhance the identification of real uneven areas, local variance is introduced as a weight to effectively suppress the interference of high-frequency discrete noise while preserving the macroscopic uniformity trend.

[0056] Specifically, the fabric uniformity of any pixel is determined based on the mean and standard deviation of the flatness index of all its neighboring pixels; the fabric uniformity satisfies the expression:

[0057]

[0058] In the formula, For the first Fabric uniformity at each pixel; For the first The mean and standard deviation of the flatness index of all neighboring pixels of a pixel; It is the maximum-minimum normalization function.

[0059] in, Reflecting the The mean of the flatness index of all neighboring pixels of a pixel, i.e. the degree of macroscopic anomaly significance; Reflecting the The weight of each pixel; if smaller and The smaller the value, the lower the flatness index in the neighborhood, indicating that the flatness index is abnormally low in value and highly clustered in spatial distribution, and is not affected by high-frequency noise. This means that the pixel is more likely to belong to the real fabric unevenness region, that is, the lower the fabric uniformity of the corresponding pixel. Furthermore, for high-frequency discrete noise in non-fabric unevenness regions, since the pixel has a larger weight, it increases the fabric uniformity of both the fabric unevenness region and the high-frequency discrete noise, ensuring the accuracy of the fabric uniformity assessment.

[0060] Furthermore, the mapped grayscale value of each pixel is calculated. , ,in For the first Fabric uniformity at each pixel; To round down, the calculated mapped grayscale value is assigned to the corresponding pixel coordinates, thereby generating a brightness distribution map of the fabric uniformity.

[0061] For example, Figure 3 This is a brightness distribution map of the uniformity of the concrete placement to be detected. The brighter areas in the image are the locations with poor uniformity, that is, the areas of abnormal accumulation or pitting of the concrete placement.

[0062] At this point, the fabric uniformity of each pixel has been obtained.

[0063] S5. Obtain the empirical threshold determined based on the distribution of qualified samples; identify pixels in the concrete placement image to be detected whose uniformity of placement is less than the empirical threshold, and mark them as non-uniform placement points.

[0064] It should be noted that setting a fixed threshold can easily lead to missed detections or false alarms when the environment changes. Therefore, by sampling different qualified samples and obtaining statistical patterns, an empirical threshold is obtained. This mechanism gives the system the ability to adapt to environmental changes and material batch differences, ensuring that it can accurately capture real fabric defects while tolerating normal process fluctuations.

[0065] Specifically, the empirical threshold is obtained by collecting multiple manually inspected and qualified concrete placement images as qualified samples, calculating the uniformity of placement of all pixels in all qualified samples and sorting them in ascending order, and selecting the value at the 5th position after sorting as the empirical threshold. This setting can ensure that abnormal accumulation features beyond the normal range are captured while tolerating normal texture fluctuations. In this embodiment of the invention, 10 qualified samples are collected, which can be adjusted according to the actual situation.

[0066] Identify pixels in the concrete placement image to be detected whose uniformity of placement is less than an empirical threshold, and mark them as uneven placement points. Calculate the ratio of the number of uneven placement points to the total number of pixels. If the ratio exceeds 3%, the current placement is determined to be uneven.

[0067] This invention also discloses a machine vision-based concrete uniformity detection system, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a machine vision-based concrete uniformity detection method according to the present invention.

[0068] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A method for detecting the uniformity of concrete placement based on machine vision, characterized in that, include: Acquire images of the concrete placement to be detected and perform smoothing processing to obtain the grayscale value and position of each pixel in the processed image; For any pixel in the processed image: based on the spatial distance between the pixel and its neighboring pixels, the absolute value of the gray level difference between the pixel and its neighboring pixels is weighted and averaged to determine the local gray level fluctuation energy. Calculate the mean and variance of the gray values ​​of all neighboring pixels of the pixel; determine the flatness index based on the local gray-level fluctuation energy of the pixel and the mean and variance; The fabric uniformity is determined based on the mean and standard deviation of the flatness index of all neighboring pixels of the given pixel. Obtain an empirical threshold determined based on the distribution of qualified samples; identify pixels in the concrete placement image to be detected whose uniformity of placement is less than the empirical threshold, and mark them as non-uniform placement points; The local grayscale fluctuation energy satisfies the expression: ; For the first Local grayscale fluctuation energy of each pixel; For the first The grayscale value of each pixel; For the first The first pixel The grayscale value of each neighboring pixel; For the first The pixel and its first Euclidean distance between neighboring pixels; This is the preset standard deviation parameter for spatial distance; These are the index values ​​and total number of neighboring pixels; To take the absolute value; It is a natural exponential function; The flatness index satisfies the following expression: ; For the first Flatness index of each pixel; For the first The mean and variance of gray levels of the neighboring pixels of a pixel; These are preset hyperparameters; The average grayscale value of all pixels; Determining fabric uniformity includes: normalizing the standard deviation of the flatness index of all neighboring pixels of each pixel to obtain the normalized standard deviation; Fabric uniformity is half the sum of the mean and normalized standard deviation of the flatness indices of all neighboring pixels of a given pixel.

2. The method for detecting the uniformity of concrete placement based on machine vision according to claim 1, characterized in that, The process of acquiring and smoothing the image of the concrete placement to be detected includes: The image of the concrete material to be detected is converted to grayscale to obtain a grayscale image; the grayscale image is then subjected to Gaussian filtering to obtain the processed image.

3. The method for detecting the uniformity of concrete placement based on machine vision according to claim 1, characterized in that, The process of obtaining the empirical threshold determined based on the distribution of qualified samples includes: Multiple manually inspected and qualified concrete placement images were collected as qualified samples; the uniformity of the placement of all pixels in all qualified samples was calculated and sorted in ascending order, and the value at the 5th percentile position after sorting was selected as the empirical threshold.

4. The method for detecting the uniformity of concrete placement based on machine vision according to claim 2, characterized in that, The Gaussian filtering process uses a filter kernel size of 1. .

5. The method for detecting the uniformity of concrete placement based on machine vision according to claim 1, characterized in that, The method further includes: obtaining data centered on each pixel. All pixels within the neighborhood window are considered as the neighboring pixels of that pixel; where, It is an odd number greater than or equal to 3.

6. The method for detecting the uniformity of concrete placement based on machine vision according to claim 1, characterized in that, The method further includes: statistically analyzing the ratio of the number of uneven points to the total number of pixels in the concrete placement image to be detected; if the ratio exceeds 3%, the current placement is determined to be uneven.

7. A machine vision-based concrete placement uniformity detection system, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a machine vision-based method for detecting the uniformity of concrete placement according to any one of claims 1-6.

Images