System and method for identifying types of stone slices and available resources

Abstract

A system and method for identifying stone slice types and available resources are provided. The system for identifying the type of stone slices and acquiring available resources includes a control and calculation circuit, a storage circuit and an image capturing device. The storage circuit stores multiple calculation models and a user interface. The image capture device is connected to the control and calculation circuit. When a stone slice is transmitted through one side of the image capture device, the image capture device captures a stone image information of the stone slice. The control and calculation circuit calculates the area value of the stone slice and determines its type classification based on at least one of a plurality of calculation models and the stone image information.

Description

[Technical Field]

[0001] This invention relates to a stone slice identification system and calculation method, and in particular to a system and method for identifying stone slice types and usable volume. [Previous Technology]

[0002] In recent years, artificial intelligence and computer vision technologies have made significant progress and have been widely applied in various traditional industries. In industry, they can identify defects and abnormal conditions, providing strong support for digital transformation and intelligent upgrading. In the marble processing industry, stone factories typically begin by slicing the raw stone, calculating its actual area, and providing customers with photos describing the slice texture and color—an indispensable part of the entire production process. However, the shape of the stone is not an ideal rectangle; its edges are irregular, often with missing or convex corners.

[0003] Traditionally, these measurement and calculation tasks were done manually, which was not only time-consuming and laborious, but also prone to discrepancies due to human subjectivity. Different manual measurement methods often led to errors, including the need to measure both the length and width each time the area of ​​a rectangle was calculated, and the deduction of missing corners varied from person to person. [Summary of the Invention]

[0004] The technical problem to be solved by the present invention is to provide a system for identifying the type of stone slice and its usable volume, which addresses the shortcomings of the prior art. The system includes: a control and calculation circuit; a storage circuit for storing multiple calculation models; and an image capturing device connected to the control and calculation circuit. When a stone slice is transmitted through one side of the image capturing device, the image capturing device captures stone image information of the stone slice. The control and calculation circuit calculates a volume value and a stone type classification of the stone slice based on the multiple calculation models and the stone image information.

[0005] This invention also discloses a method for identifying the type of stone slice and its usable volume. The stone slice calculation method includes: acquiring stone image information of a stone slice; classifying the stone slice into stone types using one of a plurality of calculation models or any combination thereof to determine the stone type classification of the stone slice; calculating the volume value of the stone slice using one of the plurality of calculation models or any combination thereof; and outputting a stone record report including a shape, a size and the volume value of each stone slice.

[0006] The system and method for identifying stone slice types and determining usable area provided by this invention can improve efficiency and reduce costs: by introducing intelligent technology and automated computing systems, labor costs and time can be significantly reduced, improving the efficiency and accuracy of stone slice area calculation. Furthermore, the system and method of this invention can also improve product quality and reliability, increasing customer satisfaction. Moreover, the system and method of this invention can promote intelligent manufacturing and digital transformation, serving as an important example of intelligent manufacturing and digital transformation in the stone industry, leading the entire stone industry towards intelligent and automated development, and enhancing industrial competitiveness and innovation capabilities.

[0007] To further understand the features and technical content of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are for reference and illustration only and are not intended to limit the present invention.

Implementation Method

[0008] The following specific embodiments illustrate the implementation of the "system and method for identifying stone slice types and usable volume" disclosed in this invention. Those skilled in the art can understand the advantages and effects of this invention from the content disclosed in this specification. This invention can be implemented or applied through other different specific embodiments, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the concept of this invention. Furthermore, the accompanying drawings of this invention are for simple illustrative purposes only and are not depictions of actual dimensions, as stated in advance. The following embodiments will further describe the relevant technical content of this invention in detail, but the disclosed content is not intended to limit the scope of protection of this invention. Additionally, the term "or" used herein should, depending on the actual situation, may include any combination of one or more of the associated listed items.

[0009] [First Embodiment]

[0010] Please refer to Figures 1, 2A, 2B, and 3. Figure 1 is a schematic diagram of the system for identifying stone slice types and usable volume according to the first embodiment of the present invention. Figure 2A is a schematic diagram of the system for identifying stone slice types and usable volume according to the first embodiment of the present invention using multiple calculation models. Figure 2B is another schematic diagram of the system for identifying stone slice types and usable volume according to the first embodiment of the present invention using multiple calculation models. Figure 3 is a schematic diagram of the stone slice of the present invention passing through a moving device.

[0011] In this embodiment, a stone slicing precision calculation system SYS1 is provided. The stone slicing precision calculation system SYS1 includes a control and calculation circuit 1, a storage circuit 2, an image capturing device 3, and a moving device 4.

[0012] Storage circuit 2 stores various calculation models. Image capturing device 3 is connected to control and calculation circuit 1. Moving device 4 is used to move stone slice S1.

[0013] When a stone slice S1 is placed on the moving device 4 and transmitted through one side of the image capturing device 3, the image capturing device 3 captures stone image information of the stone slice S1. The control and calculation circuit 1 calculates a product value and a stone type classification of the stone slice S1 based on various calculation models and the stone image information.

[0014] The control and calculation circuit 1 uses one of several calculation models to classify the stone slice S1 into different types of stone to determine the stone type classification of the stone slice S1. The control and calculation circuit 1 uses other calculation models to calculate the area value of the stone slice S1. The area value of the stone slice S1 is in units of "area". One area is an area of ​​30 cm by 30 cm.

[0015] The stone types are classified at least as follows: Italian Grey Pearl, Olive Brown, Coniferous Stone, Glass Grey, Angla Pearl, Red Travertine, Athens Grey, Carved White, Phantom of the Opera, Shell Black, Cloud Top Grey, Snowflake, Modern Grey, Old Beige, etc.

[0016] Please refer to Figure 2A. Figure 2A is a schematic diagram of the system for identifying the type of stone slices and the usable volume of stone slices according to the first embodiment of the present invention, which uses multiple calculation models for calculation.

[0017] Figure 2A shows that any one depth calculation model or multiple calculation models can be integrated for learning to classify and identify stone types and calculate the area of ​​stone image information IM. The stone image information IM can calculate the area of ​​stone slice S1 using one of the first depth calculation model M1, the second depth calculation model M2, the third depth calculation model M3, and the fourth depth calculation model M4, or any combination thereof. The stone image information IM can also classify and identify stone types using one of the first depth calculation model M1, the second depth calculation model M2, the third depth calculation model M3, and the fourth depth calculation model M4, or any combination thereof.

[0018] The Stone Image Information IM can calculate the area value of a stone slice S1 by using one of multiple depth calculation models or any combination thereof. The system in this embodiment can also classify and identify stone types by using one of multiple depth calculation models or any combination thereof.

[0019] Please refer to Figure 2B. Figure 2B is another schematic diagram of the system for identifying the type of stone slices and the usable volume of stone slices according to the first embodiment of the present invention, which uses multiple calculation models for calculation.

[0020] Figure 2B shows the different depth calculation models assigned to stone type identification and stone image information IM for volumetric calculation. Specifically, in Figure 2B, the stone image information IM calculates the volumetric value of stone slice S1 using the first depth calculation model M1, the second depth calculation model M2, and the third depth calculation model M3. The stone image information IM also classifies stone types using the fourth depth calculation model M4.

[0021] In this embodiment, taking the softmax function model as an example, it utilizes stone image information to apply a flattened layer in the image segmentation model, converts the last layer feature map of the backbone network into a one-dimensional array, and then uses the softmax function model to classify stone types. Other calculation models each have their own calculation methods to identify and classify stone types. The multiple calculation models for calculating the product value of stone slices include at least a softmax function model, a fully convolutional neural network (FCN) calculation model, a U-Net calculation model, a DeepLabV3+ calculation model, an FCN with MobileNetV2 calculation model, an FCN with ResNet50 calculation model, a U-Net with MobileNetV2 calculation model, a U-Net with ResNet50 calculation model, a DeepLabV3+ with MobileNetV2 calculation model, and a DeepLabV3+ with ResNet50 calculation model.

[0022] The method for calculating the volume of the stone slice S1, for example using the DeepLabV3+ with ResNet50 calculation model, may include the following steps:

[0023] (1) Using the weights of the DeepLabV3+ with ResNet50 model, the predicted image of the stone is obtained.

[0024] (2) Perform convex hull on the concave part of the predicted image.

[0025] (3) Detect straight lines in a graph through Hough transformation.

[0026] (4) On each line, find the point where the horizontal and vertical intersections with the point in the figure are found.

[0027] (5) Find the line with the minimum distance from the midpoint in the four directions of up, down, left, and right.

[0028] (6) By finding the four lines, intersect with the convex hull shape, and find the four innermost intersection points respectively.

[0029] (7) Draw a rectangle based on the four points obtained, and obtain the shortest length and width. The part of the fastener is obtained by subtracting 3 cm from the predicted drawing and the obtained length and width.

[0030] The above steps are just an example. In other embodiments, other different detection feature points and calculation methods can be used, which are not limited in this invention.

[0031] To ensure accurate calculation of stone slices, this embodiment also proposes an image segmentation method based on ensemble learning. This method uses FCN, U-Net, and DeepLabV3+ models, and integrates their prediction results through weighted soft voting. Soft voting is a weighted average method for calculating the probability of each model for each class. Compared to the traditional hard voting method (i.e., majority decision), soft voting is more flexible and can take into account the confidence level of each model for each pixel class, thereby improving prediction accuracy. To find the optimal weight combination, this embodiment uses a Bayesian optimization method, limiting the weight adjustment range to [0, 1]. That is, as shown in Figure 2B, multiple deep learning models are used to calculate the stone slice area, and then soft voting is performed to determine the stone slice area.

[0032] That is, different calculation models will calculate different numbers of stones and defects in stone slice S1. Moreover, for different types of stone, different calculation models may also calculate different numbers of stones and defects in stone slice S1. Therefore, in this embodiment, a soft voting method with different weights is used to calculate the stone's material volume value.

[0033] Therefore, when a stone slice is damaged, the defect value of the area of ​​the stone image information of the stone slice S1 is N times a predetermined value, where N is greater than or equal to 1. The predetermined value can be 0.5. If the defect value is less than 0.5, the defect value is calculated as 0.5. If the defect value is greater than 0.5 but less than 1, the defect value is calculated as 1. In this embodiment, the unit for calculating the area of ​​the stone image information can be adjusted according to different regions. For example, in Europe and America, feet, inches, or centimeters and meters (metric units) can be used. That is, for different regions, the system SYS1 of this embodiment can provide a predetermined value (basic unit) for the corresponding area. In other words, users can switch between regional and area calculation units through the user interface provided by this system.

[0034] Please refer to Figures 4 and 5. Figure 4 is a schematic diagram of a portable file format report of the stone record report according to the first embodiment of the present invention. Figure 5 is a schematic diagram of an Excel format report of the stone record report according to the first embodiment of the present invention.

[0035] The control and calculation circuit 1 can record the shape, size, and volume value of each stone slice S1 through the user interface (UI) to output a stone record report with a corresponding batch number. The stone record report includes a portable file format report (PDF) R1 and an Excel format report R2. That is, the user can output multiple report formats through the user interface (UI) provided by this system.

[0036] [Second Embodiment]

[0037] Please refer to Figure 6, which is a flowchart of a method for identifying the type of stone slices and the usable volume according to the second embodiment of the present invention.

[0038] In this embodiment, a method for identifying the type of stone slices and determining their usable volume is provided, comprising the following steps:

[0039] Step S601: Extract stone image information from a stone slice.

[0040] Step S602: Use one of a variety of calculation models or any combination thereof to classify the stone slices into different types of stones to determine the classification of the stone slices into different types of stones.

[0041] Step S603: Calculate the volume value of the stone slice using one of the multiple calculation models or any combination thereof.

[0042] Step S604: Output a stone record report including the shape, size and volume value of each of the stone slices.

[0043] In this embodiment, the method for identifying stone slice types and determining usable area utilizes a multi-task process. Besides calculating the area of ​​the stone slices, it simultaneously determines the stone type classification. That is, the method for identifying stone slice types and determining usable area in this embodiment can perform multiple tasks simultaneously, ultimately integrating and outputting a complete result. Furthermore, in calculating the stone area, multiple calculation models are used simultaneously, followed by a weighted soft-voting method.

[0044] In order to ensure the accurate calculation of the stone slices, the Bayesian optimization method can be used as in the first embodiment to limit the weight adjustment range to [0, 1], use multiple depth calculation models to calculate the stone slice area, and then use the soft voting method to determine the stone slice area.

[0045] Furthermore, to facilitate manufacturers in outputting stone volume data, when the stone slice S1 is damaged, the volume data value of the stone image information is a defect value that is N times a predetermined value, where N is greater than or equal to 1. The predetermined value can be 0.5 units. If the defect value is less than 0.5 units, the defect value is calculated as 0.5 units. If the defect value is greater than 0.5 units but less than 1 unit, the defect value is calculated as 1 unit.

[0046] The calculation model for classifying stone types in stone slice S1 is a softmax function model. In this embodiment, a flattened layer is used in the image segmentation model using stone image information, and the last layer feature map of the backbone network is converted into a one-dimensional array. The softmax function model is then used for stone type classification. The stone type classification includes at least Italian Grey Pearl, Olive Brown, Coniferous Stone, Glass Grey, Angla Pearl, Red Travertine, Athens Grey, Carved White, Phantom of the Opera, Shell Black, Cloud Top Grey, Snowflake, Modern Grey, and Old Beige.

[0047] The multiple calculation models for calculating the product value of the stone slices include at least a Fully Convolutional Network (FCN) calculation model, a U-Net calculation model, a DeepLabV3+ calculation model, an FCN with MobileNetV2 calculation model, an FCN with ResNet50 calculation model, a U-Net with MobileNetV2 calculation model, a U-Net with ResNet50 calculation model, a DeepLabV3+ with MobileNetV2 calculation model, and a DeepLabV3+ with ResNet50 calculation model.

[0048] [Beneficial Effects of the Embodiments]

[0049] The stone slicing calculation system and method provided by this invention can improve efficiency and reduce costs: by introducing intelligent technology and automated calculation systems, labor costs and time can be significantly reduced, and the efficiency and accuracy of stone slicing area calculation can be improved. Furthermore, the system and method of this invention can also improve product quality and reliability, and increase customer satisfaction. Moreover, the system and method of this invention can also promote intelligent manufacturing and digital transformation, serving as an important example of intelligent manufacturing and digital transformation in the stone industry, leading the entire stone industry towards intelligent and automated development, and improving industrial competitiveness and innovation capabilities.

[0050] The content disclosed above is only a preferred and feasible embodiment of the present invention, and is not intended to limit the scope of the patent application of the present invention. Therefore, all equivalent technical changes made using the contents of the present invention specification and drawings are included in the scope of the patent application of the present invention. [Simplified Explanation of the Diagram]

[0051] Figure 1 is a schematic diagram of a system for identifying the type of stone slices and the usable volume of the stone slices according to the first embodiment of the present invention.

[0052] Figure 2A is a schematic diagram of the system for identifying the type of stone slices and the usable volume of the stone slices according to the first embodiment of the present invention, which uses multiple calculation models for calculation.

[0053] Figure 2B is another schematic diagram of the system for identifying the type of stone slices and the usable volume of the stone slices according to the first embodiment of the present invention, which uses multiple calculation models for calculation.

[0054] Figure 3 is a schematic diagram of the stone slices of the present invention passing through the moving device.

[0055] Figure 4 is a schematic diagram of the portable file format report of the stone record report according to the first embodiment of the present invention.

[0056] Figure 5 is a schematic diagram of the excel format report of the stone record report of the first embodiment of the present invention.

[0057] Figure 6 is a flowchart of the method for identifying the type of stone slices and the usable volume of the slices according to the second embodiment of the present invention.

Claims

1. A system for identifying the type and usable volume of stone slabs, comprising: One control and computing circuit; A storage circuit stores multiple calculation models and a user interface; The system includes an image capturing device connected to the control and calculation circuit. When a stone slice is transmitted through one side of the image capturing device, the image capturing device captures stone image information of the stone slice. The control and calculation circuit performs convex hull processing on the concave portion of the stone image information and calculates a volume value, a defect value, and a stone type classification of the stone slice based on the various calculation models and the stone image information.

2. The system for identifying the type of stone slabs and determining the usable volume as described in claim 1, wherein, The multiple calculation models determine the weights between each of the multiple calculation models based on the pixel categories of the stone image information. The control and calculation circuit uses the multiple calculation models to classify the stone slices into different types of stones to determine the stone type classification of the stone slices. The control and calculation circuit uses the multiple calculation models to calculate the material product value of the stone slices.

3. The system for identifying the type of stone slabs and determining their usable volume as described in claim 1, wherein, The calculation model for classifying the stone slices into different types is a softmax function model. The multiple calculation models for calculating the product value of the stone slices include a Fully Convolutional Network (FCN) calculation model, a U-Net calculation model, a DeepLabV3+ calculation model, an FCN with MobileNetV2 calculation model, an FCN with ResNet50 calculation model, a U-Net with MobileNetV2 calculation model, a U-Net with ResNet50 calculation model, a DeepLabV3+ with MobileNetV2 calculation model, a softmax function, and a DeepLabV3+ with ResNet50 calculation model.

4. The system for identifying the type of stone slabs and determining their usable volume as described in claim 1, wherein, The defect value of the stone image information is N times a predetermined value, where N is greater than or equal to 1, and the predetermined value is 0.

5.

5. The system for identifying the type of stone slabs and determining the usable volume as described in claim 1, wherein, The control and calculation circuit records the shape, size, and volume value of each stone slice to output a stone record report with a corresponding batch number. The control and calculation circuit outputs the stone record report through the user interface.

6. The system for identifying stone slice types and available volume as described in claim 1, wherein the control and calculation circuit uses one of the multiple calculation models to classify the stone slices into stone types to determine the stone type classification of the stone slices, and the control and calculation circuit uses one of the multiple calculation models to calculate the volume value of the stone slices.

7. A method for identifying the type of stone slab and determining its usable volume, comprising: Extracting image information of a stone slice; The stone slices are classified into stone types and segmented into stone image information using one of a variety of calculation models or any combination thereof. Then, the concave parts of the stone image information are convex hulled according to the segmentation results, and a volume value and a defect value of the stone slices are calculated using the segmentation results. Finally, a stone record report is output, including the shape, size, volume value and defect value of each stone slice.

8. The method for identifying the type of stone slab and determining its usable volume as described in claim 7, wherein, When the stone slice is damaged, the defect value of the stone image information is N times a predetermined value, where N is greater than or equal to 1.

9. The method for identifying the type of stone slabs and determining their usable volume as described in claim 8, wherein, The predetermined value is 0.

5.

10. The method for identifying the type of stone slab and the usable volume as described in claim 9, wherein, The various computational models include a softmax function model, a fully convolutional neural network (FCN) computational model, a U-Net computational model, a DeepLabV3+ computational model, an FCN with MobileNetV2 computational model, an FCN with ResNet50 computational model, a U-Net with MobileNetV2 computational model, a U-Net with ResNet50 computational model, a DeepLabV3+ with MobileNetV2 computational model, and a DeepLabV3+ with ResNet50 computational model.

Images