Research On Caoal Dust Detection Algorithm Based On Machine Vision

Coal dust detection is a key step to achieve automatic watering and dust reduction.The main purpose of this paper is to determine whether the image contains coal dust and label it with rectangular boxes through the image of the coal operation plant area collected by the camera,especially the open coal yard.With the coal dust detected,the start and stop of the sprinkler dust reduction system can be intelligently controlled to achieve automatic watering,which can effectively control coal dust in time and save water resource.Firstly,the dataset is built by the data collected from the open-pit coal yard and the Internet.Due to the influence of the environment and the camera during the acquisition process,the image needs to be preprocessed and enhanced.In order to avoid losing the information of the coal dust itself because of excessive preprocessing,image enhancement should be focused on.In order to simulate the coal dust image under different ambient illumination,image transformation,rotation transformation and contrast brightness adjustment are used to enhance the image also enrich the dataset.The reason without adding any noise is that the image after adding noise may easily be confused with the coal dust image,which will affect the final accuracy of classification and identification.Secondly,traditional feature extraction and classification are performed on the preprocessed enhanced dataset.The local binary pattern feature descriptors are selected to describe the texture features of coal dust,and the circular local binary pattern,the rotation invariant local binary pattern,the uniform local binary pattern,and the contrast combined local binary pattern are used for feature extraction operations.Then the feature vectors are input into the support vector machine for classification.After comparing the results,the better methods are selected.Furthermore,the genetic algorithm is used to optimize the parameters of the support vector machine,which further improves the classification accuracy.Finally,the manual selection feature is omitted,and the feature is fully extracted byusing the deep learning method.Based on the classical convolutional neural network,a convolutional neural network for coal dust detection is built.The network parameters and structure are optimized and the variable learning rate is used to accelerate the convergence of the network during the training process.The model has a significant improvement in the correct rate compared to the traditional method,and can basically meet the requirements of real-time.