CN111079773A - Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network - Google Patents

Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network Download PDF

Info

Publication number
CN111079773A
CN111079773A CN201911380242.7A CN201911380242A CN111079773A CN 111079773 A CN111079773 A CN 111079773A CN 201911380242 A CN201911380242 A CN 201911380242A CN 111079773 A CN111079773 A CN 111079773A
Authority
CN
China
Prior art keywords
gravel
individual
parameter
mask
profile
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201911380242.7A
Other languages
Chinese (zh)
Inventor
郝明
王东辉
张建龙
冯兴雷
李胜伟
凌小明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chengdu Geological Survey Center Of China Geological Survey
Original Assignee
Chengdu Geological Survey Center Of China Geological Survey
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chengdu Geological Survey Center Of China Geological Survey filed Critical Chengdu Geological Survey Center Of China Geological Survey
Priority to CN201911380242.7A priority Critical patent/CN111079773A/en
Publication of CN111079773A publication Critical patent/CN111079773A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30132—Masonry; Concrete

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Geometry (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)

Abstract

The invention relates to the technical field of geological survey and exploration and the technical field of artificial intelligence, and discloses a gravel parameter acquisition method, a gravel parameter acquisition device, gravel parameter acquisition equipment and a gravel parameter storage medium based on a Mask R-CNN network. The invention provides a method for quickly, effectively and accurately carrying out intelligent recognition and parameter calculation on gravel images based on a deep learning technology, namely, on the basis of establishing a large amount of gravel photo sample libraries in the early stage, an image contour extraction model capable of intelligently recognizing gravel images can be obtained through deep learning and sample training based on a Mask R-CNN network, then, direct photographing and uploading are carried out in the later stage, artificial intelligent recognition can be carried out on gravel through the image contour extraction model, gravel parameter calculation is carried out according to individual size parameters of a reference object, scale information of gravel in the gravel images is extracted, manpower and material resources are greatly saved, deviation-free operation on parameters of single gravel is ensured, the accuracy of data is improved, and the method is convenient for practical application and popularization.

Description

Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network
Technical Field
The invention belongs to the technical field of geological survey and exploration and the technical field of artificial intelligence, and particularly relates to a gravel parameter acquisition method, a gravel parameter acquisition device, gravel parameter acquisition equipment and a gravel parameter storage medium based on a Mask R-CNN network.
Background
The study on the shape and distribution composition of the gravel plays an important role in revealing the deposition environment, the evolution of the ancient environment and the flow direction of the ancient water body, and the main study contents comprise the study on the shape (roundness), arrangement, distribution and the like of the gravel. The shape is the basic attribute of all objects including gravel, and the change of the gravel shape and the abrasion degree can enable the geographical (geological) hypothesis related to the transport distance, the water flow dynamic characteristics and the like to be expressed, but since the geological effect is a gradual process, the gravel shape is diversified, and the gravel shape is still one of the most difficult properties to describe and quantify of the objects. Whether it is the gravel shape or the gravel distribution, depends on the calculation of parameters, embodied in specific numbers, such as roundness parameters describing the shape, calculations depending on the radius of curvature, minor axis, major axis, etc., and the gravel distribution is also described depending on the specific distance between the gravel.
In field work, the traditional mathematical concept is inconvenient to operate, gravel identification needs to be carried out on a large scale, and a large amount of manpower and material resources need to be consumed. According to the current situation of gravel research at home and abroad, remote sensing images and digital images are mostly adopted for parameter inversion in large-area gravel recognition calculation, and for small-area gravel recognition, software such as AutoCAD and PS is mostly adopted for manual drawing and labeling.
Disclosure of Invention
In order to solve the problem that the existing gravel identification mode is low in efficiency and low in result reliability, and further causes difficulty in quickly and accurately acquiring gravel related parameters, the invention aims to provide a gravel parameter acquisition method, a gravel parameter acquisition device, gravel parameter acquisition equipment and a gravel parameter storage medium based on a Mask R-CNN network.
The technical scheme adopted by the invention is as follows:
a gravel parameter acquisition method based on a Mask R-CNN network comprises the following steps:
s101, introducing a plurality of gravel photo samples into a Mask R-CNN network model for training to obtain an image contour extraction model, wherein reference object individual images and gravel individual images are marked in the gravel photo samples in advance;
s102, importing the picture of the gravel to be detected containing the image of the reference object individual into the image profile extraction model to obtain the extracted profile of the reference object individual and the profile of the gravel individual;
s103, acquiring individual size parameters of the reference object corresponding to the individual outline of the reference object;
s104, aiming at each extracted individual profile of the gravel, calculating and obtaining corresponding basic parameters of the gravel individual according to the individual profile of the reference object and the individual size parameters of the reference object, wherein the basic parameters of the gravel individual comprise the number of corners, the curvature of an inscribed circle corresponding to each corner, a minor axis value, a major axis value, the width of an circumscribed rectangle, the length of the circumscribed rectangle and/or the area of the gravel.
Preferably, after the step S104, the following steps are further included:
s105, calculating to obtain the gravel roundness, the gravel flatness, the gravel sorting coefficient and/or the gravel distribution uniformity according to the basic parameters of each gravel individual in the to-be-detected gravel picture;
and S106, generating a gravel parameter chart file corresponding to the gravel photo to be detected according to the parameter calculation result.
Further optimally, in the step S105, the gravel roundness R corresponding to the profile of a single gravel individual is calculated according to the following formulaW：
Figure BDA0002342064250000021
Wherein n is the number of corners, i is a natural number between 1 and n, riR is the maximum inscribed circle curvature among the n corners for the inscribed circle curvature corresponding to the ith corner.
Further optimally, in the step S105, the gravel flatness FL corresponding to the contour of a single gravel individual is calculated according to the following formulaD：
Figure BDA0002342064250000022
In the formula, AXSIs the minor axis value, AX, of the profile of the gravel packLThe long axis value of the gravel individual profile; or, AXSCircumscribed rectangular width, AX, of the contour of the gravel packLIs the circumscribed rectangular length of the individual profile of the gravel.
Further optimally, in the step S105, the gravel separation coefficient S is calculated according to the following formula:
Figure BDA0002342064250000023
wherein N is the total number of the individual outlines of the gravels in the picture of the gravels to be detected, j is a natural number between 1 and N, and djIs the gravel diameter s corresponding to the profile of the jth individual graveljTo the gravel area corresponding to the jth gravel pack profile,
Figure BDA0002342064250000024
and the average value of the gravel diameter of all the gravel individual profiles in the to-be-tested gravel picture is obtained.
Further optimally, in the step S105, the gravel distribution uniformity is obtained as follows:
s501, dividing the gravel photo to be detected into m by adopting a grid method2Each unit cell with the same area, wherein m is a natural number between 2 and 4;
s502, for each cell, counting the total area of the gravel occupied by the corresponding individual gravel outline, and calculating the gravel diameter of the corresponding cell according to the following formula:
Figure BDA0002342064250000031
in the formula, DkRepresents the cell gravel diameter, S, corresponding to the kth cellkRepresenting the total area of gravel corresponding to the kth cell;
s503, calculating to obtain the standard deviation of the cell gravel diameter of all the cells, and taking the standard deviation of the cell gravel diameter as the gravel distribution uniformity.
The other technical scheme adopted by the invention is as follows:
a gravel parameter acquisition device based on a Mask R-CNN network comprises a model training unit, a contour extraction unit, a reference parameter acquisition unit and a basic parameter calculation unit;
the model training unit is used for introducing a plurality of gravel photo samples into a Mask R-CNN network model for training to obtain an image contour extraction model, wherein reference individual images and gravel individual images are pre-marked in the gravel photo samples;
the profile extraction unit is in communication connection with the model training unit and is used for introducing a picture of the gravel to be detected containing an image of the reference object individual into the image profile extraction model to obtain an extracted profile of the reference object individual and an extracted profile of the gravel individual;
the reference parameter acquiring unit is used for acquiring the size parameter of the individual reference object corresponding to the outline of the individual reference object;
the basic parameter calculating unit is respectively in communication connection with the profile extracting unit and the reference parameter acquiring unit and is used for calculating and obtaining corresponding basic parameters of the individual gravels according to the individual profiles of the reference objects and the individual size parameters of the reference objects aiming at the extracted individual profiles of the gravels, wherein the basic parameters of the individual gravels comprise the number of corners, the curvature of inscribed circles corresponding to the corners, the value of a short axis, the value of a long axis, the width of an circumscribed rectangle, the length of the circumscribed rectangle and/or the area of the gravels.
The optimization method further comprises a geological parameter calculation unit and a chart file generation unit;
the geological parameter calculation unit is in communication connection with the basic parameter calculation unit and is used for calculating and obtaining the gravel roundness, the gravel flatness, the gravel separation coefficient and/or the gravel distribution uniformity according to basic parameters of each gravel individual in the gravel photo to be detected;
the chart file generating unit is in communication connection with the geological parameter calculating unit and/or the basic parameter calculating unit and is used for generating a gravel parameter chart file corresponding to the gravel photo to be detected according to a parameter calculating result.
The other technical scheme adopted by the invention is as follows:
the gravel parameter acquisition equipment based on the Mask R-CNN network comprises a memory and a processor which are connected in a communication mode, wherein the memory is used for storing a computer program, and the processor is used for executing the computer program to realize the steps of the gravel parameter acquisition method based on the Mask R-CNN network.
The other technical scheme adopted by the invention is as follows:
a storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the Mask R-CNN network-based gravel parameter acquisition method as described above.
The invention has the beneficial effects that:
(1) the invention provides a method for carrying out intelligent gravel image recognition and parameter calculation based on a deep learning technology quickly, effectively and accurately, namely, on the basis of establishing a large amount of gravel photo sample libraries in the early stage, an image contour extraction model capable of intelligently recognizing gravel images can be obtained through deep learning and sample training based on a Mask R-CNN network, then, direct photographing and uploading are carried out in the later stage, artificial intelligent recognition can be carried out on gravel through the image contour extraction model, gravel parameter calculation is carried out according to individual size parameters of a reference object, scale information of gravel in the gravel images is extracted, manpower and material resources are greatly saved, deviation-free operation on parameters of single gravel is ensured, and the accuracy of data is improved;
(2) the distribution degree, the gravel separation coefficient and the like of all gravels on the picture to be detected can be further calculated, so that the geological analysis at the later stage is facilitated, the technical and data support is provided for the subsequent large environment evolution such as a gravel layer, a deposition environment, an ancient river flow direction and the like, and the practical application and popularization are facilitated;
(3) the method is very suitable for various scientific researches which need to calculate gravel parameter information in a large scale, such as researches on ancient river channels, ancient environment evolution, urban underground space utilization and the like, and compared with the traditional manual or semi-automatic calculation of gravel parameters, the method greatly improves the later processing time and has reliability, effectiveness and scientificity.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a gravel parameter acquisition method based on a Mask R-CNN network provided by the invention.
Fig. 2 is a schematic diagram of the working principle of the Mask R-CNN network provided by the present invention.
FIG. 3 is a schematic diagram of the working principle of the Faster R-CNN network provided by the present invention.
FIG. 4 is an exemplary illustration of a photograph of gravel collected in the field provided by the present invention.
FIG. 5 is an exemplary diagram of a gravel profile extracted by using a Mask R-CNN network model provided by the present invention.
FIG. 6 is a schematic structural diagram of a gravel parameter acquisition device based on a Mask R-CNN network provided by the invention.
FIG. 7 is a schematic structural diagram of a gravel parameter acquisition device based on a Mask R-CNN network provided by the invention.
Detailed Description
The invention is further described with reference to the following figures and specific embodiments. It should be noted that the description of the embodiments is provided to help understanding of the present invention, but the present invention is not limited thereto. Specific structural and functional details disclosed herein are merely illustrative of example embodiments of the invention. This invention may, however, be embodied in many alternate forms and should not be construed as limited to the embodiments set forth herein.
It will be understood that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments of the present invention.
It should be understood that, for the term "and/or" as may appear herein, it is merely an associative relationship that describes an associated object, meaning that three relationships may exist, e.g., a and/or B may mean: a exists alone, B exists alone, and A and B exist at the same time; for the term "/and" as may appear herein, which describes another associative object relationship, it means that two relationships may exist, e.g., a/and B, may mean: a exists independently, and A and B exist independently; in addition, for the character "/" that may appear herein, it generally means that the former and latter associated objects are in an "or" relationship.
It will be understood that when an element is referred to herein as being "connected," "connected," or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may be present. Conversely, if a unit is referred to herein as being "directly connected" or "directly coupled" to another unit, it is intended that no intervening units are present. In addition, other words used to describe the relationship between elements should be interpreted in a similar manner (e.g., "between … …" versus "directly between … …", "adjacent" versus "directly adjacent", etc.).
It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments of the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises," "comprising," "includes" and/or "including," when used herein, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, numbers, steps, operations, elements, components, and/or groups thereof.
It should also be noted that, in some alternative implementations, the functions/acts noted may occur out of the order noted in the figures. For example, two figures shown in succession may, in fact, be executed substantially concurrently, or the figures may sometimes be executed in the reverse order, depending upon the functionality/acts involved.
It should be understood that specific details are provided in the following description to facilitate a thorough understanding of example embodiments. However, it will be understood by those of ordinary skill in the art that the example embodiments may be practiced without these specific details. For example, systems may be shown in block diagrams in order not to obscure the examples in unnecessary detail. In other instances, well-known processes, structures and techniques may be shown without unnecessary detail in order to avoid obscuring example embodiments.
Example one
As shown in fig. 1 to 5, the gravel parameter acquiring method based on the Mask R-CNN network provided in this embodiment may include, but is not limited to, the following steps S101 to S104.
S101, introducing a plurality of gravel photo samples into a Mask R-CNN network model for training to obtain an image contour extraction model, wherein reference object individual images and gravel individual images are marked in the gravel photo samples in advance.
Before the step S101, it is necessary to obtain gravel photographs obtained according to a specific photographing rule (when the photographing rule of the gravel photographs is determined, gravel samples in different regions, such as river flood beach gravel image samples, mountain gravel image samples and flood fan gravel image samples, should be considered as much as possible to improve the diversity of the samples, and also, a photographing angle problem should be considered, when photographing, a vertical angle should be formed with respect to a gravel course, and a reference object with a fixed size, shape and color should be placed beside the gravel, so that the generated images are not easy to deform and more reliable, and the number of gravels in a single image should be considered, which can be adjusted according to actual conditions of the gravel and is not too large) so as to serve as qualified samples. Since a control object (i.e., a reference object) with fixed size, shape and color (for facilitating model identification) is added during the data acquisition phase, it can be used to calibrate the actual size of the gravel, ensuring that the subsequently extracted gravel profile is not only a qualitative but also a quantitative result. And then after the gravel photos are collected, manually arranging and analyzing the gravel photos, for example, numbering and screening the photos, screening out unqualified gravel photos, uniquely numbering the qualified gravel photos, and marking individual images of a reference object and gravel to form a field gravel photo sample library.
In the step S101, the Mask R-CNN network model is based on the existing fast R-CNN network model (R-CNN is the first algorithm for successfully applying deep learning to target detection, but some problems exist in that ① uses a selectivesearch to generate a prosal, which is time-consuming in operation and is not beneficial to the overall training and testing of the network, ② generates the prosal which needs to be subjected to warp operation and then sent to a subsequent network to cause the deformation and distortion of an image, ③ each of the propofol needs to be subjected to feature extraction independently, and has a large repeated calculation amount, in order to improve the R-CNN algorithm, a researcher proposes a FastR-CNN structure, and ROIPooling is added, so that the R-CNN algorithm is improved at three points:
Figure BDA0002342064250000063
warp operation is not needed any more, so that deformation and distortion of an object are effectively avoided, and authenticity of characteristic information is guaranteed;
Figure BDA0002342064250000062
features do not need to be extracted from each propofol, and a ROIfeature area is obtained from the feature of the whole picture in a mapping mode;
Figure BDA0002342064250000061
the Classification and regression tasks are combined into a multi-task model in the FastR-CNN, and the sharing of characteristics and the further improvement of speed are realized. However, FastR-CNN still does not solve the problems of time consumption and large storage pressure of the process of generating propofol by selectivesearch. Later, researchers put forward a FasterR-CNN algorithm, namely, a RegionProposalNet is put forward, and an RPN network shares the convolution characteristic of an input image, so that a candidate region can be generated quickly, and the calculation cost is low. Faster R-CNN can also be understood as RPN + FastR-CNN. ) The extension of the method is that the existing deep learning network model of a Mask prediction branch is added on the basis of the Faster R-CNN network model, and target detection and example segmentation can be simultaneously carried out.
The Mask R-CNN network structure is easy to realize and train, has high speed, can be conveniently applied to other fields, such as image posture estimation, character key point detection and the like, and has good effect, and the working principle is as follows: the Mask R-CNN is a combination of fast R-CNN, which is responsible for object detection (class label + window), and FCN, which is responsible for determining the target contour, as shown in FIG. 2; for each target object, FasterR-CNN has two outputs, namely a classification label and a candidate window; to segment the target pixel, a third output, a binary mask (mask) indicating the pixel location of the object in the window, may be added on top of the first two outputs. Unlike the first two outputs, this new output requires a finer spatial layout to be extracted; for this purpose, MaskR-CNN adds a branch network on the fast-RCNN: fullycondolutionnetwork (fcn), as shown in fig. 3; FCN is a popular semantic segmentation algorithm, so-called semantic segmentation, in which a machine automatically segments an object region from an image and identifies the content therein. The model first compresses the input image to the original size 1/32 through the convolutional layer and the max-pooling layer, and then performs classification prediction at this fine granularity level. Finally, it restores the graph to the original size using the upsampling and deconstruction layer. In short, it can be said that the Mask R-CNN combines two networks-FasterR-CNN and FCN are incorporated into the same jumbo fabric. The loss function of the model calculates the total loss of the classification, the generation window, and the generation mask. Therefore, multiple gravel photo samples can be trained on the basis of the existing MaskR-CNN network model, and the trained MaskR-CNN network model can be used as the image contour extraction model by adjusting the model parameters in the training process.
And S102, importing the picture of the gravel to be detected containing the image of the reference object individual into the image profile extraction model to obtain the extracted profile of the reference object individual and the profile of the gravel individual.
Before the step S102, the gravel picture to be detected also needs to be acquired according to a specific shooting rule of the gravel picture sample, and the used reference object is also the same reference object, so as to identify the reference object individual and the gravel individual through the image contour extraction model. As shown in fig. 4, in the photograph of the gravel to be measured, a regular circular plate (specifically, red color) is the reference object. As also shown in FIG. 5, gravel individual contours can be extracted by using the trained MaskR-CNN network model.
S103, acquiring the individual size parameter of the reference object corresponding to the individual contour of the reference object.
Before the step S103, the individual size parameters of the reference object are pre-bound and stored in a storage unit.
S104, calculating corresponding basic parameters of the gravel individual according to the extracted individual profiles of the reference objects and the individual size parameters of the reference objects, wherein the basic parameters of the gravel individual can include, but are not limited to, the number of corners, the curvature of an inscribed circle corresponding to each corner, a minor axis value, a major axis value, the width of an circumscribed rectangle, the length of the circumscribed rectangle, the area of gravel and the like.
In the step S104, since the reference object has a fixed size and shape, after the size parameters of the individual size of the reference object are obtained, the basic parameters of the individual gravel of each individual gravel profile can be calculated by using a conventional algorithm based on the size comparison result of the individual profile of the reference object and the individual gravel profile, so that surveying and mapping personnel are not required to participate in the measurement and calculation process of the gravel parameters, and only direct photographing and uploading are required at a later stage, so that artificial intelligent identification and parameter calculation can be performed on the gravel, manpower and material resources are greatly saved, non-deviation calculation on the parameters of the single gravel is ensured, the accuracy of data is improved, and practical application and popularization are facilitated.
In order to further acquire more detailed gravel parameters and obtain corresponding data reports, the following steps S105 to S106 may be further included after the step S104.
And S105, calculating the roundness of the gravel, the flatness of the gravel, the separation coefficient of the gravel and/or the distribution uniformity of the gravel according to the basic parameters of each gravel individual in the picture of the gravel to be detected.
In the step S105, the gravel roundness degree is a ratio of an average curvature radius of the gravel angle to a maximum inscribed circle radius, and the gravel roundness degree R corresponding to the individual profile of a single gravel can be calculated according to the following formulaW：
Figure BDA0002342064250000081
Wherein n is the number of corners, i is a natural number between 1 and n, riR is the maximum inscribed circle curvature among the n corners for the inscribed circle curvature corresponding to the ith corner.
In the step S105, the gravel flatness FL corresponding to the individual gravel contour can be calculated according to the following formulaD：
Figure BDA0002342064250000082
In the formula, AXSIs the minor axis value, AX, of the profile of the gravel packLThe long axis value of the gravel individual profile; or, AXSCircumscribed rectangular width, AX, of the contour of the gravel packLIs the circumscribed rectangular length of the individual profile of the gravel.
In the step S105, the gravel separation coefficient indicates the distribution degree of the gravel sizes, and the gravel separation coefficient S can be calculated according to the following formula:
Figure BDA0002342064250000083
wherein N is the total number of the individual outlines of the gravels in the picture of the gravels to be detected, j is a natural number between 1 and N, and djIs the gravel diameter s corresponding to the profile of the jth individual graveljTo the gravel area corresponding to the jth gravel pack profile,
Figure BDA0002342064250000084
and the average value of the gravel diameter of all the gravel individual profiles in the to-be-tested gravel picture is obtained.
In the step S105, the gravel distribution uniformity mainly reflects the difference in the gravel quantity in the to-be-measured gravel picture, and the gravel distribution uniformity can be obtained specifically as follows: s501, dividing the gravel photo to be detected into m by adopting a grid method2Each unit cell with the same area, wherein m is a natural number between 2 and 4; s502, for each cell, counting the total area of the gravel occupied by the corresponding individual gravel outline, and calculating the gravel diameter of the corresponding cell according to the following formula:
Figure BDA0002342064250000091
in the formula, DkRepresents the cell gravel diameter, S, corresponding to the kth cellkRepresenting the total area of gravel corresponding to the kth cell; s503, counting to obtain the standard deviation of the cell gravel diameters of all the cellsThe standard deviation of the unit cell gravel diameter is used as the gravel distribution uniformity.
And S106, generating a gravel parameter chart file corresponding to the gravel photo to be detected according to the parameter calculation result.
In the step S106, the parameter calculation result includes the gravel individual basic parameter, the gravel roundness degree, the gravel flatness degree, the gravel separation coefficient, and/or the gravel distribution uniformity, and the like. In addition, the specific details and procedures for generating the gravel parameter graph file can be realized by referring to the conventional manner, such as counting and drawing by a computer program, and obtaining various graph files required by researchers. Therefore, through the steps S105-S106, the distribution degree of all gravels, the gravel separation coefficient and the like on the picture to be detected can be further calculated, the geological analysis at the later stage is facilitated, and the technology and data support are provided for the evolution of large environments such as the subsequent gravel layer, the deposition environment, the ancient river flow direction and the like.
In summary, the gravel parameter obtaining method based on Mask R-CNN network provided by this embodiment has the following technical effects:
(1) the embodiment provides a method for quickly, effectively and accurately carrying out intelligent gravel image identification and parameter calculation based on a deep learning technology, namely, on the basis of establishing a large amount of gravel photo sample libraries in the early stage, an image contour extraction model capable of intelligently identifying gravel images can be obtained through deep learning and sample training based on a Mask R-CNN network, then, direct photographing and uploading are carried out in the later stage, artificial intelligent gravel identification can be carried out through the image contour extraction model, gravel parameter calculation is carried out according to individual size parameters of a reference object, scale information of gravel in the gravel images is extracted, manpower and material resources are greatly saved, deviation-free operation on parameters of single gravel is ensured, and the accuracy of data is improved;
(2) the distribution degree, the gravel separation coefficient and the like of all gravels on the picture to be detected can be further calculated, so that the geological analysis at the later stage is facilitated, the technical and data support is provided for the subsequent large environment evolution such as a gravel layer, a deposition environment, an ancient river flow direction and the like, and the practical application and popularization are facilitated;
(3) the method is very suitable for various scientific researches which need to calculate gravel parameter information in a large scale, such as researches on ancient river channels, ancient environment evolution, urban underground space utilization and the like, and compared with the traditional manual or semi-automatic calculation of gravel parameters, the method greatly improves the later processing time and has reliability, effectiveness and scientificity.
Example two
As shown in fig. 6, the present embodiment provides a hardware apparatus for implementing the gravel parameter obtaining method based on the Mask R-CNN network according to the first embodiment, and the hardware apparatus includes a model training unit, a contour extraction unit, a reference parameter obtaining unit, and a basic parameter calculating unit; the model training unit is used for introducing a plurality of gravel photo samples into a Mask R-CNN network model for training to obtain an image contour extraction model, wherein reference individual images and gravel individual images are pre-marked in the gravel photo samples; the profile extraction unit is in communication connection with the model training unit and is used for introducing a picture of the gravel to be detected containing an image of the reference object individual into the image profile extraction model to obtain an extracted profile of the reference object individual and an extracted profile of the gravel individual; the reference parameter acquiring unit is used for acquiring the size parameter of the individual reference object corresponding to the outline of the individual reference object; the basic parameter calculating unit is respectively in communication connection with the profile extracting unit and the reference parameter acquiring unit and is used for calculating and obtaining corresponding basic parameters of the individual gravels according to the individual profiles of the reference objects and the individual size parameters of the reference objects aiming at the extracted individual profiles of the gravels, wherein the basic parameters of the individual gravels comprise the number of corners, the curvature of inscribed circles corresponding to the corners, the value of a short axis, the value of a long axis, the width of an circumscribed rectangle, the length of the circumscribed rectangle and/or the area of the gravels.
The optimization method further comprises a geological parameter calculation unit and a chart file generation unit; the geological parameter calculation unit is in communication connection with the basic parameter calculation unit and is used for calculating and obtaining the gravel roundness, the gravel flatness, the gravel separation coefficient and/or the gravel distribution uniformity according to basic parameters of each gravel individual in the gravel photo to be detected; the chart file generating unit is in communication connection with the geological parameter calculating unit and/or the basic parameter calculating unit and is used for generating a gravel parameter chart file corresponding to the gravel photo to be detected according to a parameter calculating result.
The working process, working details and technical effects of the foregoing apparatus provided in this embodiment may be referred to in the first embodiment, and are not described herein again.
EXAMPLE III
As shown in fig. 7, this embodiment provides a hardware device for implementing the gravel parameter acquiring method based on the Mask R-CNN network according to the first embodiment, and the hardware device includes a memory and a processor, which are communicatively connected, where the memory is used to store a computer program, and the processor is used to execute the computer program to implement the gravel parameter acquiring method based on the Mask R-CNN network according to the first embodiment.
The working process, the working details and the technical effects of the foregoing device provided in this embodiment may be referred to as embodiment one, and are not described herein again.
Example four
The present embodiment provides a storage medium storing a computer program including the Mask R-CNN network-based gravel parameter obtaining method according to the first embodiment, that is, a computer program is stored on the storage medium, and when the computer program is executed by a processor, the steps of the Mask R-CNN network-based gravel parameter obtaining method according to the first embodiment are implemented. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices, or may be a mobile intelligent device (such as a smart phone, a PAD, or an ipad).
For the working process, the working details, and the technical effects of the storage medium provided in this embodiment, reference may be made to embodiment one, which is not described herein again.
The various embodiments described above are merely illustrative, and may or may not be physically separate, as they relate to elements illustrated as separate components; if reference is made to a component displayed as a unit, it may or may not be a physical unit, and may be located in one place or distributed over a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: modifications of the technical solutions described in the embodiments or equivalent replacements of some technical features may still be made. And such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Finally, it should be noted that the present invention is not limited to the above alternative embodiments, and that various other forms of products can be obtained by anyone in light of the present invention. The above detailed description should not be taken as limiting the scope of the invention, which is defined in the claims, and which the description is intended to be interpreted accordingly.

Claims (10)

1. A gravel parameter acquisition method based on a Mask R-CNN network is characterized by comprising the following steps:
s101, introducing a plurality of gravel photo samples into a Mask R-CNN network model for training to obtain an image contour extraction model, wherein reference object individual images and gravel individual images are marked in the gravel photo samples in advance;
s102, importing the picture of the gravel to be detected containing the image of the reference object individual into the image profile extraction model to obtain the extracted profile of the reference object individual and the profile of the gravel individual;
s103, acquiring individual size parameters of the reference object corresponding to the individual outline of the reference object;
s104, aiming at each extracted individual profile of the gravel, calculating and obtaining corresponding basic parameters of the gravel individual according to the individual profile of the reference object and the individual size parameters of the reference object, wherein the basic parameters of the gravel individual comprise the number of corners, the curvature of an inscribed circle corresponding to each corner, a minor axis value, a major axis value, the width of an circumscribed rectangle, the length of the circumscribed rectangle and/or the area of the gravel.
2. The method for acquiring the gravel parameters based on the Mask R-CNN network as claimed in claim 1, further comprising the following steps after the step S104:
s105, calculating to obtain the gravel roundness, the gravel flatness, the gravel sorting coefficient and/or the gravel distribution uniformity according to the basic parameters of each gravel individual in the to-be-detected gravel picture;
and S106, generating a gravel parameter chart file corresponding to the gravel photo to be detected according to the parameter calculation result.
3. The method for acquiring the gravel parameters based on the Mask R-CNN network as claimed in claim 2, wherein in the step S105, the gravel roundness R corresponding to the profile of a single gravel individual is calculated according to the following formulaW：
Figure FDA0002342064240000011
Wherein n is the number of corners, i is a natural number between 1 and n, riR is the maximum inscribed circle curvature among the n corners for the inscribed circle curvature corresponding to the ith corner.
4. The method for acquiring gravel parameters based on Mask R-CNN network as claimed in claim 2, wherein in step S105, the gravel flatness FL corresponding to the individual profile of the single gravel is calculated according to the following formulaD：
Figure FDA0002342064240000012
In the formula, AXSIs gravelMinor axis value of body profile, AXLThe long axis value of the gravel individual profile; or, AXSCircumscribed rectangular width, AX, of the contour of the gravel packLIs the circumscribed rectangular length of the individual profile of the gravel.
5. The method for acquiring the gravel parameters based on the Mask R-CNN network as claimed in claim 2, wherein in the step S105, the gravel separation coefficient S is calculated according to the following formula:
Figure FDA0002342064240000021
wherein N is the total number of the individual outlines of the gravels in the picture of the gravels to be detected, j is a natural number between 1 and N, and djIs the gravel diameter s corresponding to the profile of the jth individual graveljTo the gravel area corresponding to the jth gravel pack profile,
Figure FDA0002342064240000022
and the average value of the gravel diameter of all the gravel individual profiles in the to-be-tested gravel picture is obtained.
6. The method for acquiring gravel parameters based on Mask R-CNN network as claimed in claim 2, wherein in step S105, the gravel distribution uniformity is obtained as follows:
s501, dividing the gravel photo to be detected into m by adopting a grid method2Each unit cell with the same area, wherein m is a natural number between 2 and 4;
s502, for each cell, counting the total area of the gravel occupied by the corresponding individual gravel outline, and calculating the gravel diameter of the corresponding cell according to the following formula:
Figure FDA0002342064240000023
in the formula, DkRepresents the cell gravel diameter, S, corresponding to the kth cellkRepresenting the total area of gravel corresponding to the kth cell;
s503, calculating to obtain the standard deviation of the cell gravel diameter of all the cells, and taking the standard deviation of the cell gravel diameter as the gravel distribution uniformity.
7. A gravel parameter acquisition device based on a Mask R-CNN network is characterized by comprising a model training unit, a contour extraction unit, a reference parameter acquisition unit and a basic parameter calculation unit;
the model training unit is used for introducing a plurality of gravel photo samples into a Mask R-CNN network model for training to obtain an image contour extraction model, wherein reference individual images and gravel individual images are pre-marked in the gravel photo samples;
the profile extraction unit is in communication connection with the model training unit and is used for introducing a picture of the gravel to be detected containing an image of the reference object individual into the image profile extraction model to obtain an extracted profile of the reference object individual and an extracted profile of the gravel individual;
the reference parameter acquiring unit is used for acquiring the size parameter of the individual reference object corresponding to the outline of the individual reference object;
the basic parameter calculating unit is respectively in communication connection with the profile extracting unit and the reference parameter acquiring unit and is used for calculating and obtaining corresponding basic parameters of the individual gravels according to the individual profiles of the reference objects and the individual size parameters of the reference objects aiming at the extracted individual profiles of the gravels, wherein the basic parameters of the individual gravels comprise the number of corners, the curvature of inscribed circles corresponding to the corners, the value of a short axis, the value of a long axis, the width of an circumscribed rectangle, the length of the circumscribed rectangle and/or the area of the gravels.
8. The gravel parameter acquisition device based on the Mask R-CNN network as claimed in claim 7, further comprising a geological parameter calculation unit and a chart file generation unit;
the geological parameter calculation unit is in communication connection with the basic parameter calculation unit and is used for calculating and obtaining the gravel roundness, the gravel flatness, the gravel separation coefficient and/or the gravel distribution uniformity according to basic parameters of each gravel individual in the gravel photo to be detected;
the chart file generating unit is in communication connection with the geological parameter calculating unit and/or the basic parameter calculating unit and is used for generating a gravel parameter chart file corresponding to the gravel photo to be detected according to a parameter calculating result.
9. A gravel parameter acquisition equipment based on Mask R-CNN network is characterized in that: the gravel parameter acquisition method based on the Mask R-CNN network comprises a storage and a processor which are in communication connection, wherein the storage is used for storing a computer program, and the processor is used for executing the computer program to realize the gravel parameter acquisition method steps based on the Mask R-CNN network according to any one of claims 1-6.
10. A storage medium, characterized by: the storage medium stores a computer program, and the computer program is executed by a processor to realize the steps of the gravel parameter acquisition method based on the Mask R-CNN network according to any one of claims 1-6.
CN201911380242.7A 2019-12-27 2019-12-27 Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network Pending CN111079773A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911380242.7A CN111079773A (en) 2019-12-27 2019-12-27 Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911380242.7A CN111079773A (en) 2019-12-27 2019-12-27 Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network

Publications (1)

Publication Number Publication Date
CN111079773A true CN111079773A (en) 2020-04-28

Family

ID=70318797

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911380242.7A Pending CN111079773A (en) 2019-12-27 2019-12-27 Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network

Country Status (1)

Country Link
CN (1) CN111079773A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112153320A (en) * 2020-09-23 2020-12-29 北京京东振世信息技术有限公司 Method and device for measuring size of article, electronic equipment and storage medium
CN112528913A (en) * 2020-12-18 2021-03-19 中山艾尚智同信息科技有限公司 Grit particulate matter particle size detection analytic system based on image
CN113137225A (en) * 2021-04-12 2021-07-20 中国地质调查局油气资源调查中心 Gravel reservoir evaluation method based on gravel component analysis
CN113177949A (en) * 2021-04-16 2021-07-27 中南大学 Large-size rock particle feature identification method and device
CN115562292A (en) * 2022-10-24 2023-01-03 广州市南电电力工程有限公司 Comprehensive control system for cable pipeline threading robot

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110263835A (en) * 2019-06-13 2019-09-20 中国电建集团华东勘测设计研究院有限公司 Rock category automatic identifying method based on deep learning and Bayesian network

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110263835A (en) * 2019-06-13 2019-09-20 中国电建集团华东勘测设计研究院有限公司 Rock category automatic identifying method based on deep learning and Bayesian network

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
KAIMING HE等: "《Mask R-CNN》", 《2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION(ICCV)》 *
杨光: "《岩心图像砾石分析技术在砂砾岩扇体中的应用》", 《油气地质与采收率》 *
穆悦: "《戈壁表面多尺度砾石特征参数估算及其空间分布规律研究》", 《中国博士学位论文全文数据库 基础科学辑》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112153320A (en) * 2020-09-23 2020-12-29 北京京东振世信息技术有限公司 Method and device for measuring size of article, electronic equipment and storage medium
CN112528913A (en) * 2020-12-18 2021-03-19 中山艾尚智同信息科技有限公司 Grit particulate matter particle size detection analytic system based on image
CN113137225A (en) * 2021-04-12 2021-07-20 中国地质调查局油气资源调查中心 Gravel reservoir evaluation method based on gravel component analysis
CN113177949A (en) * 2021-04-16 2021-07-27 中南大学 Large-size rock particle feature identification method and device
CN113177949B (en) * 2021-04-16 2023-09-01 中南大学 Large-size rock particle feature recognition method and device
CN115562292A (en) * 2022-10-24 2023-01-03 广州市南电电力工程有限公司 Comprehensive control system for cable pipeline threading robot
CN115562292B (en) * 2022-10-24 2023-07-18 广州市南电电力工程有限公司 Comprehensive control system for cable duct threading robot

Similar Documents

Publication Publication Date Title
CN111079773A (en) Gravel parameter acquisition method, device, equipment and storage medium based on Mask R-CNN network
CN111460984B (en) Global lane line detection method based on key points and gradient equalization loss
CN109145830B (en) Intelligent water gauge identification method
CN108921166A (en) Medical bill class text detection recognition method and system based on deep neural network
CN107844802A (en) Water and soil conservation value method based on unmanned plane low-altitude remote sensing and object oriented classification
CN108346144A (en) Bridge Crack based on computer vision monitoring and recognition methods automatically
CN103324938A (en) Method for training attitude classifier and object classifier and method and device for detecting objects
CN111310674B (en) Remote sensing monitoring method and device for river ecological buffer zone based on grid division technology
CN109284779A (en) Object detecting method based on the full convolutional network of depth
CN108564085A (en) A kind of method of automatic reading pointer type instrument reading
CN111914767A (en) Scattered-pollution enterprise detection method and system based on multi-source remote sensing data
CN108596881A (en) The intelligent image statistical method of rock grain size after a kind of explosion
CN108664970A (en) A kind of fast target detection method, electronic equipment, storage medium and system
CN109165666A (en) Multi-tag image classification method, device, equipment and storage medium
CN110197113A (en) A kind of method for detecting human face of high-precision anchor point matching strategy
CN111767826A (en) Timing fixed-point scene abnormity detection method
CN110427506A (en) Spatial data boundary processing method, device, computer equipment and storage medium
CN117523573A (en) Water meter reading method and system based on multi-stage visual detection fusion fault early warning
CN116403071B (en) Method and device for detecting few-sample concrete defects based on feature reconstruction
CN110135422B (en) Dense target detection method and device
CN113762153B (en) Novel tailing pond detection method and system based on remote sensing data
CN113642582B (en) Ammeter reading identification method and device, electronic equipment and storage medium
CN113239815B (en) Remote sensing image classification method, device and equipment based on real semantic full-network learning
CN108399413B (en) Picture shooting area identification and geographical positioning method and device
CN112699824B (en) Method and device for detecting constant of electric energy meter and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20200428