Blast heap ore rock particle image segmentation method

An image segmentation, ore-rock technology, applied in the field of image processing, can solve problems such as serious adhesion of ore-rock particles, inaccurate contour extraction, complex parameter adjustment, etc., to avoid over-segmentation and under-segmentation, good light adaptive performance, and segmentation high precision effect

Pending Publication Date: 2019-06-07
NORTHEASTERN UNIV
View PDF3 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problems of serious adhesion of ore-rock particles, inaccurate contour extraction, poor segmentation effect and complex parameter adjustment existing in the existing image segmentation technology, the present invention provides a method for image segmentation of explosive-heap ore-rock particles

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Blast heap ore rock particle image segmentation method
  • Blast heap ore rock particle image segmentation method
  • Blast heap ore rock particle image segmentation method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0048] Such as Figure 4 As shown, the invention provides a method for image segmentation of explosive pile ore rock particles, specifically comprising the following steps:

[0049] As shown in 1, obtaining the pre-trained first convolutional neural network and the second convolutional neural network includes:

[0050] A1, constructing the first convolutional neural network and the second convolutional neural network;

[0051] Such as figure 2 As shown, for example, the first convolutional neural network is a U-Net convolutional neural network whose input image size is 48×48; as image 3 The second convolutional neural network shown is a U-Net convolutional neural network with an input image size of 480×480. The U-Net convolutional network includes a 3×3 convolutional layer, a maximum pooling layer, and a 2×2 upsampling volume Convolutional Layers and 1×1 Convolutional Layers

[0052] The U-Net convolutional network adopted in the present invention is based on a deformati...

Embodiment 2

[0077] In the present embodiment, for any stope acquisition format is an image of explosive pile ore rock particles in RGB, such as Figure 5 As shown; it includes images with uneven illumination, complex surface texture of ore rock particles, and insufficient contrast. In addition, since the original image is in RGB format, this will increase the computational complexity in network training, so it needs to be equal to the original two-dimensional image Perform grayscale and bilateral filtering.

[0078] For the first time, put it into the first convolutional network model with an input size of 48×48 pre-training to get the first probability map, such as Figure 6 Shown; Use OpenCV to extract the contour and get the first contour map and calculate the S value of the contour map.

[0079] Generate the nth contour map for any nth probability map, use the nth contour map to modify the first probability map to obtain the nth probability correction map to correct the probability m...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention relates to a blast heap ore rock particle image segmentation method. The method comprises the following steps: S1, inputting a to-be-segmented ore rock particle image of the blasting muck pile into a pre-trained first convolutional neural network model to obtain a first probability graph and a first profile graph; S2, correcting the first probability graph by using the perimeter andthe area of the closed contour to obtain a first corrected probability graph; S3, inputting the first correction probability graph into a pre-trained second convolutional neural network model to obtain a second probability graph and a second contour graph; S4, obtaining an (n + 1) th probability graph by means of a second convolutional neural network until the resolution ratio of the total area ofall contours in the (n + 1) th contour graph to the first probability graph is greater than or equal to 1; S5, adding the nth contour map and the (n + 1) th contour map to serve as a segmentation map; according to the method, the contour is extracted by adopting the convolutional network of deep learning, the continuity is good, the number of miscellaneous points is small, the edges of the adhered ore rock particles are accurately segmented, adjustment of complex parameters is avoided, the segmentation precision is high, and the practicability is good.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for segmenting images of blast pile ore rock particles. Background technique [0002] The block size of blasting ore rock is not only the main index to verify the blasting process, but also affects the production efficiency, operating cost, and even production safety of the mine. Therefore, the real-time and accurate segmentation of the blasting ore rock particle image is very important for mine production. has important meaning. However, due to the complex distribution of explosive pile rocks, irregular shapes, and serious mutual occlusion of piled rocks, it has brought great difficulties to the segmentation of ore rock particle images. It is not suitable for the segmentation of ore particle images; [0003] At present, the ore particle images mainly use OTSU, cluster analysis, watershed and its improved methods, and image-based segmentation algorithms. Such ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06T7/00G06T7/12
CPCY02P90/30
Inventor 柳小波张育维赵胜
Owner NORTHEASTERN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap