Method and device for analyzing particle size distribution in fractions of a rock mass

Abstract

The invention relates to the mining industry. A fragmentation analysis device for determining volumetric particle size distribution in fractions of a rock mass includes a main module capable of: generating a depth map of an image using three modules for match evaluation, direction evaluation and post-processing, respectively; calculating the linear dimensions of segments, taking into account the depth map data; calculating the volumes of segments in the analysis region; and generating a fractional particle size distribution analysis on the basis of the calculated volumes and the linear dimensions of the segments. The device contains a 2D video processor, a 3D accelerator, an internal memory, a display, a power subsystem, an external data storage medium, a group of sensors, a left stereoscopic camera, and a right stereoscopic camera, and is designed to be capable of using neural network algorithms. The device is transportable and portable.

Description

[0001] METHOD AND DEVICE FOR ANALYZING THE GRANULOMETRIC COMPOSITION OF ROCK MASS FRACTIONS

[0002] DESCRIPTION

[0003] Field of technology to which the invention relates

[0004] The present invention is intended for determining and analyzing the fragmentation of the volumetric granulometric composition of fractions of the surface of a rock mass formed, for example, as a result of drilling and blasting operations in quarries of mining enterprises. In particular, the present invention relates to a method for determining the granulometric volumetric composition of fractions of a rock mass and to a device for determining the granulometric volumetric composition of fractions of a rock mass.

[0005] Quarry safety requirements prohibit approaching the quarry wall at close range, so it is necessary to determine the sizes of rock fragments from a safe distance. The working range of distances for stereo photography is from 3 to 30 m.

[0006] The developed device for determining the granulometric volumetric composition of rock mass fractions is a portable device with an autonomous power supply, operating in the "offline" mode without the Internet, and within the framework of the implementation of the method for determining the granulometric volumetric composition of rock mass fractions has the following functions:

[0007] • photographing the surface of the rock mass directly in the quarry with the stereo images linked to a geolocation point; automatic detection of individual segments in the images obtained during the shooting using neural network algorithms and classification of all segments by type: void, small / medium / large fraction, rock fragments; calculation of the volumes of all rock fragments in the stereo image and calculation of the volume of the small / medium / large fraction based on the segments identified in the image;

[0008] • analysis of fragmentation of the volumetric granulometric composition of fractions of the surface of the rock mass directly on the device itself, without the need to connect to remote computing and cloud resources (offline mode).

[0009] The device for determining the fragmentation of the granulometric volumetric composition of rock mass fractions automatically determines the distances to the fragments being removed and, in accordance with the distances, effectively calculates their actual linear dimensions.

[0010] The device for determining the fragmentation of the granulometric volumetric composition of rock mass fractions performs automatic granulometric segmentation to isolate individual solid fragments (pieces) of the rock mass and automatically classifies the isolated segments by the following types: solid pieces of rock mass; "small stuff" areas - consisting of small particles, the size of which is lower than the resolution of the cameras to isolate them into individual segments; "void" areas - areas that, due to insufficient lighting, cannot be visually or automatically classified by fragment type, as well as areas containing foreign objects that are not subject to consideration in the calculation of the fractional granulometric composition.

[0011] The device for determining the fragmentation of the granulometric volumetric composition of rock mass fractions provides an analysis of the fragmentation of the granulometric composition with the minimum required linear size of the measured particles on the surface area of ​​the rock mass. The device determines the minimum linear size of a piece from 1 cm at a distance of 10 m. The analysis takes no more than 30 seconds from the moment of shooting.

[0012] The results of the analysis are displayed on the device screen, in particular, in the form of a graph of the distribution of the volumes of particle content depending on their linear dimensions, and / or in the form of a histogram reflecting the maximum sizes of pieces for intervals of size fractions. The size interval of all particles of the analyzed substance is divided into classes (fractions) and the granulometric composition is presented, in particular, in the form of a percentage volume content of particles of each of the fractions (fractional composition).

[0013] The results of the analysis are presented, in particular, in the form of a report that can be sent by e-mail and / or transferred to a personal computer (PC). The analysis of the fragmentation of the granulometric composition is carried out automatically, without the need to use standards during the survey.

[0014] State of the art

[0015] A method for determining the fragmentation of the granulometric composition of the rock mass collapse is known from the prior art [RU2388998, publication date: 20.10.2009]. When performing the method, an arbitrarily oriented scaling rectangle with arbitrarily selected side lengths is placed in any accessible place, the rock mass collapse is photographed at any angle, the photoplanogram is entered into a computer, a quadrangular calculation contour of an arbitrary size and position is formed on it, the areas of the rock mass pieces are outlined, size classes are set within the calculation contour, and using a computer program, the granulometric composition of the crushed rock mass is calculated by assigning the areas of the outlined rock mass pieces to the area of ​​the calculation contour.

[0016] The disadvantage of the known technical solution is low safety, a large amount of time spent and very low accuracy in determining the fragmentation of the granulometric composition of the rock mass collapse due to the need to use a scaling rectangle during photography, which is ineffective at small angles of inclination of the collapse due to the length of the photography scene in perspective, as well as the need for personnel to be in dangerous areas of the collapse when placing the scaling rectangle, which significantly reduces the technological effectiveness of the method.

[0017] Disclosure of invention

[0018] The objective of the present invention is to eliminate the above-mentioned disadvantages of the prior art. Namely, the objective of the present invention is to create a new, safe, fast, accurate, reliable portable device with an autonomous power supply for determining the fragmentation of the granulometric volumetric composition of rock mass fractions and a method based on it, capable of operating highly efficiently without Internet access (in "offline" mode).

[0019] The technical result of using the present invention is:

[0020] 1) in improving the safety of work based on the results of drilling and blasting operations to analyze the fragmentation of the granulometric composition in quarries of mining enterprises; 2) in improving the accuracy and reliability and automation of the determination and analysis of the fragmentation of the volumetric granulometric composition of fractions of the surface of the rock mass;

[0021] 3) in increasing the speed of determination and analysis of fragmentation of the volumetric granulometric composition of fractions of the surface of the rock mass;

[0022] 4) in simplifying the determination and analysis of fragmentation of the volumetric granulometric composition of fractions of the surface of the rock mass through the use of a portable device;

[0023] 5) in expanding the arsenal of technical means for determining and analyzing the fragmentation of the volumetric granulometric composition of fractions of the surface of the rock mass;

[0024] 6) in increasing the level of security of the information received and analyzed on a portable innovative high-tech device, since the entire technology based on artificial intelligence neural networks works highly efficiently WITHOUT THE INTERNET in “offline” mode.

[0025] The above technical result is achieved by a method for determining the fragmentation of the granulometric volumetric composition of rock mass fractions, which includes the following stages: performing stereo photography and forming a stereo image; processing the image with the determination of rock segments; forming a depth map of the stereo image; calculating the linear dimensions of the segments taking into account the obtained depth map data; calculating the volumes of the segments in the area of ​​fragmentation analysis of the granulometric composition; forming a fractional granulometric analysis based on the calculated volumes of the segments and their linear dimensions; and displaying the fractional granulometric analysis on the screen of a portable handheld device.According to the present invention, at the stage of shooting and forming a stereo image, two stereo cameras are used, left and right, frames from the left and right stereo cameras are captured synchronously at the time of shooting, for both frames, correction of geometric distortions and correction of the relative position of the stereo cameras are performed.

[0026] According to the present invention, after the stage of shooting and forming a stereo image, the analysis area can be further determined and adjusted.

[0027] According to the present invention, at the stage of image processing with segment determination, the original image is processed using three neural network models: a model for identifying void areas, a model for identifying small fraction areas, and a model for identifying rock fragments, and the segments are classified by type: void, small / medium / large fraction, rock fragment.

[0028] According to the present invention, users can additionally perform necessary manual adjustments to the segments.

[0029] According to the present invention, the stage of forming a depth map of a stereo image is carried out by means of three modules: coincidence assessment, direction assessment and post-processing.

[0030] According to the present invention, at the stage of calculating the linear dimensions of a segment, the linear size of the segment is calculated as the distance between two maximally distant points of the segment, wherein the distance calculation is performed in three-dimensional space taking into account information from the depth map.

[0031] Thus, in order to achieve the above-mentioned technical result, a device is proposed for determining the fragmentation of the granulometric volumetric composition of rock mass fractions, including a main module containing a 2D-operations video processor, a 3D-accelerator, internal memory, displays, a power subsystem, an external data storage, a group of sensors, left and right stereo cameras and made with the possibility of using neural network algorithms.

[0032] According to the present innovative high-tech invention, the device may also additionally contain a sound system and a communication module.

[0033] Brief description of the drawings

[0034] The present invention is illustrated by the following drawing figures.

[0035] Fig. 1 shows a comparison of pixels during a centrally symmetric transformation of the left and right images when assessing the coincidence within the framework of the implementation of the semi-global matching algorithm (stage of forming an image depth map);

[0036] Fig. 2 shows a segment, its maximum length, and a bounding rectangle (at the stage of determining the linear dimensions of segments);

[0037] Fig. 3 demonstrates the division of a segment into stripes for calculating volume (stage of calculating segment volumes);

[0038] Fig. 4 shows a tabular representation of the results of fragmentation of granulometric analysis;

[0039] Fig. 5 shows a graphical representation of the results of fragmentation of granulometric analysis;

[0040] Fig. 6 shows the results of the granulometric analysis fragmentation in the form of a histogram;

[0041] Fig. 7 shows the display of the auxiliary table of the general distribution of fragmentation of the granulometric composition;

[0042] Fig. 8 shows the architecture of the hardware of the claimed portable device; Fig. 9 shows the architecture of the software of the claimed portable device.

[0043] Implementation of the invention

[0044] The method for determining the fragmentation of the granulometric volumetric composition of fractions of rock mass according to the present invention in its most general form contains the following steps, some of which are optional.

[0045] The operating procedure of the device consists of the following steps:

[0046] 1. Shooting and forming a stereo image.

[0047] 2. Definition and adjustment of the analysis area.

[0048] 3. Image processing with segment detection.

[0049] 4. Possible manual adjustment of segments.

[0050] 5. Formation of a depth map of the stereo image.

[0051] 6. Calculation of linear dimensions of segments taking into account depth map data.

[0052] 7. Calculation of the volumes of segments (fragments) in the analysis area.

[0053] 8. Formation of fractional granulometric analysis based on the calculated volumes of segments and their linear dimensions.

[0054] 9. Output of fractional granulometric analysis in the form of graphs and tables on the device screen and in the form of reports.

[0055] 10. Transfer reports via e-mail, to external media or to a PC.

[0056] 11. Direct analysis of the obtained data by the user with subsequent optimization of drilling and blasting processes.

[0057] The stage of shooting and forming a stereo image.

[0058] The image is formed in the following steps: 1. The operator, focusing on the image from the left camera of the device for determining the fragmentation of the granulometric volumetric composition of rock mass fractions, selects an angle and takes a picture.

[0059] 2. Frames from the left and right stereo cameras are captured synchronously at the moment of shooting.

[0060] 3. For both frames, correction of geometric distortions and correction of the relative position of the stereo cameras are performed.

[0061] 4. Based on the frames from the left and right stereo cameras, a depth map of the image is constructed, which is then used to calculate the linear dimensions, distances and volume of segments.

[0062] 5. The operator can adjust the analysis area on the image displayed on the device display.

[0063] 6. Next, the left frame is passed to the ML (Machine Learning) module to extract segments.

[0064] To construct a depth map, one of the algorithms related to semi-global matching algorithms is used, for example, the SGBM (Semi-Global Block Matching) algorithm.

[0065] This algorithm processes images from the left and right stereo cameras (after distortion correction).

[0066] The implementation of such an algorithm consists of three main modules: match evaluation, direction evaluation, and post-processing.

[0067] In the matching evaluation, the calculation is performed in two steps: the center-symmetric census transform (CSCT) of the left and right images and the Hamming distance calculation. First, the algorithm calculates the CSCT value for all pixels of the left and right images separately using a sliding window. For the current pixel, a 9 by 7 pixel window is selected around it. The CSCT value for the center pixel in this window is estimated by comparing the value of each pixel with its corresponding center-symmetric counterpart in the window. If the pixel value is greater than the corresponding center-symmetric pixel, the result is 1, otherwise the result is 0. An example of a 9 by 7 window is shown in Fig. 1. The center pixel number is 31. The 0th pixel is compared with the 62nd pixel (blue), the 1st pixel is compared with the 61st pixel (red), and so on, to obtain 31 results.Each result is output as a single bit, and the result of the entire window is formatted as a 31-bit number. This 31-bit number represents the CSCT output for each pixel in both images.

[0068] The next step (Hamming distance calculation) determines the bitwise difference between the CSTC values ​​for the corresponding pixels of the left and right images. The Hamming distance shows the number of different bits for two 31-bit numbers.

[0069] To calculate the match score, the Hamming distances are calculated between the pixels of the left image and several pixels of the right image that follow the current pixel to the right in accordance with the specified distance parameter D. The match score consists of all calculated Hamming distance values. If necessary, the list of values ​​for the match score can be adjusted for a specified number of depth levels.

[0070] The matching cost is not calculated for the pixel positions corresponding to the first columns D of the left image.

[0071] The second module of such an algorithm is the direction estimation. Due to noise in the original images, the result of the matching estimation may be ambiguous, and some incorrect matches may have a lower cost than correct ones. Therefore, additional constraints are needed to improve smoothness by penalizing changes in neighboring differences. This constraint is implemented by combining one-dimensional paths with the minimum cost from several directions. It is represented by the cumulative cost in several directions at each pixel position.

[0072] The post-processing module selects the most probable values ​​for the depth map, smooths outliers, and ensures that the depth map selects values ​​with minimal discrepancies.

[0073] The stage of image processing with segmentation (segmentation of fragments in the image).

[0074] Segmentation is the selection of areas on the original image that belong to one fragment of rock.

[0075] Rock fragments are highlighted in images of two scales:

[0076] • entire images reduced to a size multiple of 512 pixels;

[0077] • sections of the original image measuring 512 by 512 pixels without additional resizing.

[0078] The results of the sseeggmmeennttaatsii obtained from different scales are combined in the subsequent steps of the algorithm.

[0079] The original image is processed using three neural network models:

[0080] • model for highlighting areas with voids (areas of the sky, foreign objects);

[0081] • model for highlighting areas of fine fraction;

[0082] • model for identifying rock fragments of medium and large fractions.

[0083] These models are standard CNN (convolutional neural network) architectures for image segmentation tasks, when it is necessary not only to determine the class of the image as a whole, but also to segment its regions by class, i.e. to create a mask that will divide the image into several classes. The architecture consists of a contracting path for capturing context and a symmetric expanding path that allows for precise localization. In the device, this model is used to separate regions with rock fragments, small / medium / large fractions, and voids.

[0084] The model for extracting rock fragments is a unified platform for training models for object detection, instance segmentation, and image classification. The device uses this model to extract rock fragments from an image.

[0085] Also at the segmentation stage, the segments are classified by type: void, small / medium / large fraction, rock fragment.

[0086] The stage of calculating (determining) the linear dimensions of segments.

[0087] Based on the depth map data, the main module calculates the physical dimensions of each pixel for each image, which are stored in a special data structure called a pixel size map. The "pixel size" parameter shows how many millimeters of real distance each pixel occupies in the processed image.

[0088] When processed in the ML module, all segments in the images are marked as pieces, areas of small items and areas of emptiness, in which the illumination is insufficient for recognition or where unrecognizable foreign objects are located.

[0089] Since a rock fragment may be partially covered by other fragments, segments are formed based on only the visible portion of the fragments.

[0090] Each segment is characterized by a linear size in centimeters, which is taken to be equal to the maximum length of the fragment. The linear size of the segment is calculated as the distance between two maximally distant points of the segment. It is taken into account that a fragment of rock is a three-dimensional object and the distance is calculated in three-dimensional space taking into account information from the depth map.

[0091] After determining the most distant points from each other, a bounding rectangle is constructed around the segment, in which the linear size of the rock fragment is equal to the long side of the rectangle (see Fig. 2).

[0092] If individual segments were changed during editing, the linear dimensions of the segments and their volumes are recalculated.

[0093] The stage of calculating the volumes of segments in the analysis area.

[0094] The volume of a segment (fragment) is calculated as the volume of a body of revolution. The axis of rotation is taken to be a straight line passing through the two most distant points of the fragment boundary. The image of the fragment is rotated so that this axis lies horizontally.

[0095] To determine the total volume of the small fraction, it is first necessary to determine in the image the ratio of the area of ​​the small fraction segments to the area of ​​the rock fragment segments.

[0096] The volume of the small-sized fraction is calculated based on the assumption that the ratio of the volumes of the small-sized fraction and the volumes of whole fragments coincides with the ratio of the areas of the corresponding segments.

[0097] To calculate areas in an image, each pixel that is part of a fragment or detail is found. The area occupied by this pixel is found as h * w, where h is the height (the distance in millimeters between the pixels above and below the one being considered), w is the width (the distance in millimeters between the pixels to the left and right of the one being considered).

[0098] Next, the total area of ​​the small areas and the total area of ​​all segments are calculated. Since calculating the area for each pixel requires a large amount of computing resources, the calculation is performed for every 10th pixel vertically and horizontally. Tests have shown that with this approach, the value changes insignificantly for the final result, and the calculation is completed much faster.

[0099] Calculation of distribution of ppoo fractions, small, working and oversized fractions and average piece.

[0100] In the application settings, the user specifies:

[0101] • Maximum size of small fractions;

[0102] • Minimum size of oversized fragments.

[0103] According to the given parameters, the segments are divided into 3 categories:

[0104] • undersized - fragments with linear dimensions smaller than those established for the working range;

[0105] • fragments of the working range;

[0106] • oversized items – fragments with linear dimensions greater than those established for the working range.

[0107] The average piece size is also calculated.

[0108] The formula for the volume of rotation calculates the volumes of all segments in the image analysis area. This volume is summed up and taken as 100%.

[0109] From it, the volume distribution is calculated as a percentage by ten fractions (in 10% increments) of fragments depending on their maximum linear size for each fraction.

[0110] The volume distribution in percentage of fractions is also calculated for 10 bins with specified ranges of linear dimensions of pieces for each bin.

[0111] Displaying the analysis results (on the screen, in the form of a report).

[0112] The results of granulometric analysis fragmentation can be displayed as a diagram or a histogram. When displaying the results of granulometric analysis as a diagram, 2 artifacts are displayed:

[0113] 1) Tabular representation;

[0114] 2) Graphic display.

[0115] The table view (Fig. 4) displays 2 columns:

[0116] • P-value;

[0117] • Maximum particle size.

[0118] The “P-value” column displays designations from P-10 to P-100, indicating the number of the fraction of 10% of the total sample volume.

[0119] The “Maximum particle size” column contains the maximum linear size of fragments in centimeters from the range of values ​​that fall within the volume of the selected fraction.

[0120] In the graphical display (Fig. 5), each point of the graph is a separate fraction. The X axis displays the maximum linear size of pieces in each fraction. The Y axis displays the percentage of the total volume for each fraction. The minimum and maximum linear size for the working range are marked separately with vertical lines.

[0121] When displaying the analysis results in the form of a histogram, the distribution by bunkers is displayed in two types (Fig. 6):

[0122] • tabular presentation;

[0123] • histogram.

[0124] The table representation contains 10 rows to display 10 bins. Each bin corresponds to a range of linear fragment sizes, which is determined based on the specified "bin step" parameter or manually for each bin. For example, if a step of 20 cm is specified, then the first bin (numbered 0) contains a fraction of fragments with linear dimensions from 0 to 20 cm, the next one - from 20 cm to 40 cm, etc.

[0125] The table representation consists of 4 columns: • Bunker number

[0126] • Size – maximum linear size of fragments in cm for each bin.

[0127] • Range yield - the volume percentage of fragments that fall into this bin

[0128] • Output is more than - volumetric percentage of fragments entering subsequent bins.

[0129] The total volume of fragments is taken to be 100%. Then, for each bin, the percentage of the total volume of fragments with linear dimensions corresponding to the range of the corresponding bin is calculated.

[0130] The "Output More Than" column displays the percentage of fragments that belong to subsequent bins.

[0131] Example: For the initial bin, the "Yield in Range" column is 66.2% - this is the percentage of the volume of all fragments that fell into this bin. Then the "Yield More Than" column is 33.8%, since this is the percentage of the remaining stones that will be distributed among the remaining bins. For the next bin, with an "Yield in Range" of 10%, the "Yield More Than" value is: 33.8 - 10 = 23.8%.

[0132] Additionally, the auxiliary table “General distribution” is displayed (Fig. 7).

[0133] This table shows the percentage of the total fragment volume that is occupied by oversized fragments of the particle size distribution, the working range, and small fragments of the particle size distribution. It also shows the average size of all fragments, weighted by volume. When displaying the histogram, each column corresponds to a separate bin, the height of the column corresponds to the values ​​of the "Output in range" column. The graph line on top of the histogram is plotted based on the values ​​of the "Output more than" column.

[0134] Transfer of granulometric analysis results in the form of reports. Reports can be generated in the following formats:

[0135] • PDF;

[0136] • CSV;

[0137] • XLSX.

[0138] Data transfer is implemented in 2 ways:

[0139] • sending the generated report by e-mail;

[0140] • connection to a PC via cable (the device is displayed in the system as a flash drive) and manual transfer of generated reports using the PC operating system, as well as by using portable storage devices (USB drive).

[0141] Hardware architecture.

[0142] Fig. 8 shows a block diagram of a device for determining the fragmentation of the granulometric volumetric composition of rock mass fractions.

[0143] The main module includes:

[0144] • 2D operations video processor - to optimize operations with raster data (images from cameras, depth maps, sizes, area and volume of segments);

[0145] • 3D accelerator / ML block - for hardware acceleration of neural network algorithms; • internal memory - non-volatile memory for storing images together with accompanying information, generated reports.

[0146] The display is designed for visual control of the shooting area during shooting, viewing and adjusting the segmentation results, viewing the analysis results and generated reports.

[0147] The power subsystem (battery, charger) provides the device with electric power at different voltage levels (for different components), energy saving, control and management of the battery charge during charging.

[0148] External data storage (SD card) is designed to increase the amount of memory available for storing information, as well as for transferring files to a PC.

[0149] Sensors (GPS, illumination, compass, gyroscope) provide the device with the necessary information (illumination level, geographic coordinates, position of the device in space).

[0150] A sound system and a communication module (Wi-Fi, LTE modem) may also be provided.

[0151] Software architecture.

[0152] The high-level software architecture (Fig. 9) includes the following modules:

[0153] • calibration - is responsible for determining the actual parameters of the cameras and stereo-optical distortions introduced by differences in individual stereo cameras;

[0154] • image capture - obtaining stereo images from the left and right cameras; • image correction - correcting distortions arising from differences in the actual parameters of the cameras and their optical distortions;

[0155] • image processing - includes the construction of a depth map, image segmentation, calculation of the size and volume of fragments;

[0156] • data storage - ensures the storage of images, depth maps, metadata about segments, metadata about shooting parameters (illumination, device position, geographic coordinates, etc.) and reports;

[0157] • GUI (Graphical User Interface) – interaction with the user;

[0158] • libraries - libraries included in the application that are used by different modules. Highlighted separately, since one library can be used by several modules.

Claims

CLAUSE OF THE INVENTION 1. A method for determining the fragmentation of the granulometric volumetric composition of rock mass fractions, including the following stages: - carrying out shooting and forming a stereo image; - image processing with segment detection; - formation of a depth map of the image using three modules: coincidence assessment, direction assessment and post-processing; - calculation of linear dimensions of segments taking into account depth map data; - calculation of segment volumes in the analysis area; - formation of fractional granulometric analysis based on the calculated volumes of segments and their linear dimensions; and output of fractional granulometric analysis to the screen of a portable device.

2. The method according to paragraph 1, in which at the stage of shooting and forming a stereo image, two stereo cameras are used, left and right, frames from the left and right stereo cameras are captured synchronously at the moment of shooting, and for both frames, correction of geometric distortions and correction of the relative position of the stereo cameras are performed.

3. The method according to paragraph 1, in which after the stage of shooting and forming a stereo image, the area of analysis is additionally determined and adjusted.

4. The method according to paragraph 1, wherein at the stage of image processing with segment determination, the initial image is processed using three neural network models: a model for identifying areas with voids, a model for identifying areas of small fractions, and a model for identifying fragments of medium and large rocks, and the segments are classified by type: void, small / medium / large fraction, rock fragment.

5. The method according to paragraph 1, in which manual adjustment of the segments is additionally performed.

6. The method according to paragraph 1, in which, at the stage of calculating the linear dimensions of a segment, the linear dimension of a segment is calculated as the distance between two points of the segment that are most distant from each other, and the distance is calculated in three-dimensional space taking into account information from the depth map.

7. A device for determining the fragmentation of the granulometric volumetric composition of rock mass fractions for implementing the method according to any of paragraphs 1-6, including a main module configured to form an image depth map by means of three modules: assessing coincidence, assessing directions and post-processing, calculating the linear dimensions of segments taking into account the depth map data, calculating the volumes of segments in the analysis area and forming a fractional granulometric analysis based on the calculated volumes of segments and their linear dimensions, and containing a 2D operations video processor, a 3D accelerator, internal memory, displays, a power subsystem, an external data storage, a group of sensors, left and right stereo cameras, and configured to use neural network algorithms, while it is portable and handheld.

8. The device according to paragraph 7, additionally containing a sound system and a communication module.

Images