阅读说明：本技术 基于激光成像的人工智能矿石识别装置和方法 (Artificial intelligence ore recognition device and method based on laser imaging ) 是由 李建春 于 2020-06-24 设计创作，主要内容包括：本发明公开了基于激光成像的人工智能矿石识别装置和方法,涉及光电检测、光电成像和光电测量技术领域,能够对矿石进行精确识别,实时性和适用性较高。其中装置包括激光照射模块、信号传输模块、成像模块、信息识别模块。激光照射模块用于发射激光束照射传送带上的待识别矿石,并获取待识别矿石的返回信号处理得到矿石的数据信息。待识别矿石的数据信息经信号传输模块传输至成像模块。成像模块根据待识别矿石的数据信息对待识别矿石进行三维成像,得到待识别矿石的纹理信息和外形信息,送入信息识别模块。信息识别模块,对待识别矿石的纹理信息和外形信息进行归一化处理,并使用卷积神经网络对归一化处理后的信息进行识别,输出矿石识别结果。(The invention discloses an artificial intelligence ore identification device and method based on laser imaging, relates to the technical field of photoelectric detection, photoelectric imaging and photoelectric measurement, can accurately identify ores, and is high in instantaneity and applicability. The device comprises a laser irradiation module, a signal transmission module, an imaging module and an information identification module. The laser irradiation module is used for emitting laser beams to irradiate the ore to be identified on the conveyor belt and acquiring the return signals of the ore to be identified to process the return signals to obtain the data information of the ore. And the data information of the ore to be identified is transmitted to the imaging module through the signal transmission module. The imaging module carries out three-dimensional imaging on the ore to be identified according to the data information of the ore to be identified, obtains texture information and appearance information of the ore to be identified, and sends the texture information and the appearance information into the information identification module. And the information identification module is used for carrying out normalization processing on the texture information and the appearance information of the ore to be identified, identifying the information after the normalization processing by using a convolutional neural network and outputting an ore identification result.)

1. The artificial intelligent ore identification device based on laser imaging is characterized by comprising a laser irradiation module, a signal transmission module, an imaging module and an information identification module;

the laser irradiation module is used for emitting laser beams to irradiate the ore to be identified on the conveyor belt and acquiring the return signals of the ore to be identified to process so as to obtain the data information of the ore;

the data information of the ore to be identified is transmitted to the imaging module through the signal transmission module;

the imaging module carries out three-dimensional imaging on the ore to be identified according to the data information of the ore to be identified to obtain texture information and appearance information of the ore to be identified, and the texture information and the appearance information are sent to the information identification module;

and the information identification module is used for carrying out normalization processing on the texture information and the appearance information of the ore to be identified, identifying the information after the normalization processing by using a convolutional neural network and outputting an ore identification result.

2. The apparatus of claim 1, wherein the laser illumination module comprises a lidar and a pair of optoelectronic switches;

the photoelectric switch pair comprises two photoelectric switches which are oppositely arranged on two side edges of the conveyor belt and form linear correlation;

when the ore to be identified passes through the linear correlation position of the photoelectric switch pair, the linear correlation of the photoelectric switch pair is interrupted, and the photoelectric switch pair sends a trigger signal to the laser radar;

after receiving the trigger signal, the laser radar starts to emit laser to irradiate the ore to be identified on the conveyor belt;

and the laser radar receives the laser return signal reflected by the ore to be identified, and performs data conversion to obtain point cloud data of the ore to be identified as data information of the ore to be identified.

3. The apparatus of claim 1, wherein the laser illumination module comprises a laser emitter and a camera;

the laser emitter is arranged right above the conveyor belt; the laser transmitter emits laser beams to irradiate ores to be identified on the conveyor belt;

the camera is arranged above the conveyor belt in an inclined mode, shoots laser beams irradiated on ores to be identified, shoots the laser beams, and takes shot video data as data information of the ores to be identified.

4. The device according to claim 3, wherein the imaging module performs three-dimensional imaging on the ore to be identified according to the data information of the ore to be identified, specifically:

the data information of the ore to be identified comprises video data of the laser beam irradiation process on the ore, and the three-dimensional imaging result of the ore to be identified is obtained through synthesis reduction according to the change condition of the laser beam on the ore at each moment in the video data.

5. The artificial intelligent ore identification method based on laser imaging is characterized by comprising the following steps:

s1, irradiating the ore to be identified by adopting a laser beam, and acquiring a return signal of the ore to be identified to process to obtain data information of the ore;

s2, performing three-dimensional imaging on the ore to be identified according to the data information of the ore to obtain texture information and appearance information of the ore to be identified;

and S3, performing normalization processing on the texture information and the shape information of the ore to be recognized, recognizing the information after the normalization processing by using a convolutional neural network, and outputting the recognition result of the ore to be recognized.

6. The method according to claim 5, wherein the step of irradiating the ore to be identified with the laser beam and obtaining the return signal of the ore to be identified is performed to obtain data information of the ore to be identified, and specifically comprises:

emitting laser beams by using a laser radar to irradiate the ores to be identified;

and the laser radar receives an echo signal returned by the ore to be identified, and performs data conversion on the echo signal to obtain point cloud data of the ore to be identified as data information of the ore to be identified.

7. The method according to claim 5, wherein the irradiating the ore with the laser beam and obtaining the return signal of the ore to be identified are processed to obtain data information of the ore to be identified, specifically:

and irradiating the laser beam on the ore to be identified, and acquiring video data of the irradiation process of the laser beam on the ore to be identified as data information of the ore to be identified.

8. The method according to claim 7, characterized in that the ore to be identified is imaged in three dimensions according to the data information of the ore to be identified, specifically:

and obtaining a three-dimensional imaging result of the ore to be identified through synthetic reduction according to the change condition of the laser beam on the ore to be identified at each moment in the video data.

9. The method according to claims 5 to 8, characterized in that the information after the normalization processing is identified by using a convolutional neural network, and the identification result of the ore to be identified is output, specifically:

constructing a convolutional neural network model;

for the ores of the known category, the method of steps S1 and S2 is adopted to obtain texture information and shape information of the ores of the known category and perform normalization processing, the normalization processing result of the ores of the known category is utilized to train the convolutional neural network model, the normalization processing information of the ores to be recognized is input to the trained convolutional neural network model, and the recognition result is output.

Technical Field

The invention relates to the technical field of photoelectric detection, photoelectric imaging and photoelectric measurement, in particular to an artificial intelligent ore identification device and method based on laser imaging.

Background

Based on the current situation, coal mines, coal preparation plants and mines have the separation requirements of coal blocks/ores and coal gangue/miscellaneous stones. Taking coal washing processing as an example, the amount of gangue discharged per 1 hundred million tons of coking coal washed is 2000 more than ten thousand tons. At present, the large gangue screening mainly adopts manual sorting, so that the efficiency is low, the management difficulty and the labor intensity are high, and the occupational diseases are caused at the same time.

At present, imaging identification methods in the ore field mainly comprise a high-definition camera imaging method, an infrared camera imaging method and an X-ray imaging method. The high-definition camera imaging method comprises the steps of shooting ores on a conveyor belt through a camera, then carrying out a series of information processing on ore information, extracting characteristics such as gray scale, texture and appearance of the ores, and finally classifying the ores by using a classification algorithm or a convolutional neural network.

Due to the severe working environment, the radiation hazard, the identification accuracy, the real-time performance and the like, the application effect of the methods in ore identification is poor.

Therefore, there is a need for an ore identification technology with high real-time performance and high applicability.

Disclosure of Invention

In view of this, the invention provides an artificial intelligence ore identification device and method based on laser imaging, which can accurately identify ores and have high real-time performance and applicability.

In order to achieve the purpose, the technical scheme of the invention is as follows: artificial intelligence ore recognition device based on laser formation of image is including laser irradiation module, signal transmission module, imaging module, information identification module.

The laser irradiation module is used for emitting laser beams to irradiate the ore to be identified on the conveyor belt and acquiring the return signals of the ore to be identified to process the return signals to obtain the data information of the ore.

And the data information of the ore to be identified is transmitted to the imaging module through the signal transmission module.

The imaging module carries out three-dimensional imaging on the ore to be identified according to the data information of the ore to be identified, obtains texture information and appearance information of the ore to be identified, and sends the texture information and the appearance information into the information identification module.

And the information identification module is used for carrying out normalization processing on the texture information and the appearance information of the ore to be identified, identifying the information after the normalization processing by using a convolutional neural network and outputting an ore identification result.

Further, the laser irradiation module comprises a laser radar and a photoelectric switch pair.

The photoelectric switch pair comprises two photoelectric switches which are oppositely arranged on two side edges of the conveyor belt and form straight line correlation.

When the ore to be identified passes through the linear correlation position of the photoelectric switch pair, the linear correlation of the photoelectric switch pair is interrupted, and the photoelectric switch pair sends a trigger signal to the laser radar.

And after receiving the trigger signal, the laser radar starts to emit laser to irradiate the ore to be identified on the conveyor belt.

And the laser radar receives the laser return signal reflected by the ore to be identified, and performs data conversion to obtain point cloud data of the ore to be identified as data information of the ore to be identified.

Further, the laser irradiation module comprises a laser emitter and a camera; the laser emitter is arranged right above the conveyor belt; the laser emitter emits laser beams to irradiate ores to be identified on the conveyor belt;

the camera is arranged obliquely above the conveyor belt, shoots the laser beam irradiated on the ore to be identified, and takes the shot video data as the data information of the ore to be identified.

Further, the imaging module carries out three-dimensional imaging on the ore to be identified according to the data information of the ore to be identified, and specifically comprises the following steps: the data information of the ore to be identified comprises video data of the laser beam irradiation process on the ore, and the three-dimensional imaging result of the ore to be identified is obtained through synthesis reduction according to the change condition of the laser beam on the ore at each moment in the video data.

Another embodiment of the present invention provides an artificial intelligence ore identification method based on laser imaging, which is characterized by comprising the following steps:

and S1, irradiating the ore to be identified by using a laser beam, and acquiring a return signal of the ore to be identified to process to obtain data information of the ore.

And S2, performing three-dimensional imaging on the ore to be identified according to the data information of the ore to obtain texture information and appearance information of the ore to be identified.

And S3, performing normalization processing on the texture information and the shape information of the ore to be recognized, recognizing the information after the normalization processing by using a convolutional neural network, and outputting the recognition result of the ore to be recognized.

Further, adopt the laser beam to shine the ore of waiting to discern to obtain the data information that the return signal processing of waiting to discern the ore obtained and wait to discern the ore, specifically do: emitting a laser beam by using a laser radar to irradiate the ore to be identified; and receiving an echo signal returned by the ore to be identified by the laser radar, and performing data conversion on the echo signal to obtain point cloud data of the ore to be identified as data information of the ore to be identified.

Further, adopt the laser beam to shine the ore to obtain the data information that the return signal processing of waiting to discern the ore obtained and wait to discern the ore, specifically be: and irradiating the laser beam on the ore to be identified, and acquiring video data of the irradiation process of the laser beam on the ore to be identified as data information of the ore to be identified.

Further, three-dimensional imaging is carried out on the ore to be identified according to the data information of the ore to be identified, and the three-dimensional imaging method specifically comprises the following steps: and obtaining a three-dimensional imaging result of the ore to be identified through synthetic reduction according to the change condition of the laser beam on the ore to be identified at each moment in the video data.

Further, the information after the normalization processing is identified by using a convolutional neural network, and the identification result of the ore to be identified is output, specifically:

and constructing a convolutional neural network model.

For the ores of the known category, the method of steps S1 and S2 is adopted to obtain texture information and shape information of the ores of the known category and normalize the texture information and the shape information, the normalized result of the ores of the known category is used to train the convolutional neural network model, the normalized information of the ores to be recognized is input to the trained convolutional neural network model, and the recognition result is output.

Has the advantages that:

according to the artificial intelligent ore identification device and method based on laser imaging, provided by the embodiment of the invention, the ore is identified by using laser, the identification precision can reach the millimeter level, the surface texture, the appearance and the whole volume of the ore can be identified, and the texture and appearance characteristics of the ore surface cannot be lost due to the fact that the ore surface is covered by dust. The working process is independent of ambient light, and the anti-vibration device has strong anti-vibration capability and explosion-proof property. And the identification of the ore texture and the appearance information is more accurate by utilizing the convolutional neural network model. And the operation process is simple, the applicability is good, and the ore on the conveying belt can be identified in real time. Therefore, the method has higher real-time performance and practicability in the colleagues who realize the accurate identification of the ores.

Drawings

Fig. 1 is a schematic diagram of an artificial intelligence ore identification device based on laser imaging according to an embodiment of the present invention;

fig. 2 is a schematic diagram of an artificial intelligence ore identification device based on laser imaging according to another embodiment of the present invention;

FIG. 3 is a flowchart of an artificial intelligence ore identification method based on laser imaging according to an embodiment of the present invention

Detailed Description

The invention is described in detail below by way of example with reference to the accompanying drawings.

The invention provides an artificial intelligence ore recognition device based on laser imaging, which comprises a laser irradiation module, a signal transmission module, an imaging module and an information recognition module.

The laser irradiation module is used for emitting laser beams to irradiate the ore to be identified on the conveyor belt and acquiring the return signals of the ore to be identified to process the return signals to obtain the data information of the ore.

And the data information of the ore to be identified is transmitted to the imaging module through the signal transmission module.

The imaging module carries out three-dimensional imaging on the ore to be identified according to the data information of the ore to be identified, obtains texture information and appearance information of the ore to be identified, and sends the texture information and the appearance information into the information identification module.

And the information identification module is used for carrying out normalization processing on the texture information and the appearance information of the ore to be identified, identifying the information after the normalization processing by using a convolutional neural network and outputting an ore identification result.

The ore identification device provided by the invention comprises the following two embodiments: