Rapid detection method for ultralimit particles of irregular particle swarm based on machine vision

A technology of machine vision and detection methods, applied in neural learning methods, instruments, computer parts, etc., can solve problems such as high algorithm complexity, inability to meet online granularity detection, and cumbersome algorithms

CN114863277APending Publication Date: 2022-08-05INNER MONGOLIA UNIV OF TECH +1
3 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2022-08-05

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses a machine vision-based irregular particle swarm over-limit particle rapid detection method, which comprises the following steps of: collecting particle flow pictures with different illuminance and different contents in advance as a data set, performing real-time over-limit particle granularity detection by using an industrial camera, rapidly identifying a range region in which the over-limit particles are located through YOLOv5, and determining whether the over-limit particles are over-limit particles or not. And segmenting the target region through a watershed algorithm to obtain an overrun particle edge, and calculating the area of the overrun particle edge to judge whether the overrun particle is an overrun particle. According to the method, the ore group pictures with different illuminance and different granularities are collected as the training set, so that the range of the ultralimit particles can be quickly identified and extracted. A watershed algorithm is further used to segment the image in the range to obtain an ultralimit particle contour, and particle size area information is further obtained. According to the method of combining area determination through the neural network and contour segmentation through the watershed algorithm, the speed of detecting the ultralimit particles in real time in industrial application can be increased, the accuracy rate is high, and the real-time performance of detection is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention relates to the technical field of particle size detection, in particular to a rapid detection method for particle group out-of-limit particles based on machine vision. Background technique

[0002] The ore particle size parameter can be used as the main standard to measure the degree of ore crushing. At the same time, it is also the main reference for influencing the selection of beneficiation methods and technological processes. When it is detected that out-of-limit particles exist in the selected raw materials, it will affect the sorting effect of the main equipment and cause a series of hazards, such as possibly blocking the feeding pump, feeding pipeline and underflow nozzle. The upper limit rate of the product is also one of the main indicators for judging the quality of coal. In actual production scenarios, it is often encountered that the grading sieve plate is damaged or dropped, the crusher is worn, etc., resulting in e...

Examples

Embodiment Construction

[0033] The technical solution of the present invention will be further described in detail below through an embodiment (scenario of raw coal feed transportation in a coal preparation plant) and in conjunction with the accompanying drawings.

[0034] 1. Construction of coal flow images in the training set: In order to ensure the overall performance of the subsequent deep learning algorithm and avoid over-fitting, the image collection needs to consider the graphical information of coal flows under different lighting (natural light, lighting) conditions and different particle size ranges. , in order to achieve accurate extraction of granularity parameters, a known distance reference object (bar scale) is set in the image acquisition area.

[0035]2. The environment of the belt in the coal preparation plant is dark, so that the coal flow picture does not have high resolution and clear identification conditions. Industrial cameras are affected by many factors such as dust and light...