Design And Realization On Coal Mine Safety Hidden Trouble Management And Assessment System

According to the latest statistics,hearing disability in our country is the first of the five serious disability,such as vision disability,physical disability and mental disability,for 20.57 million,accounting for 1.67% of the country's population,800 thousand of whom were under the age of 7 years old children.Sign language plays an irreplaceable role as a necessary tool for deaf mute communication and an important bridge to communicate with people who listen to it and it is also one of the hot topics in the University laboratory.In this paper,Leap Motion,the latest somatosensory controller developed by Leap Corporation,is used as hand gesture feature acquisition device,aiming at the study of static finger language development.The paper has designed two kinds of finger language recognition methods;one is based on template matching method,which is based on feature matching for finger language recognition.This method extracts the feature from the spatial relation of Leap Motion and combines the shape feature of the specific finger as the feature of the finger,then the template threshold is calculated and the matching is realized.The experimental results show that this method has better adaptability to complex background,but it is difficult to extend the gesture library.The other is a finger language recognition based on support vector machines(SVM),which is mainly a statistical classification method based on the shortcomings of previous methods.This method optimizes the features extracted before,so that the extracted gesture features have good universality.After normalization and principal component analysis(PCA)dimension reduction,these features improve the system running speed,and then improve the performance of the classifier through grid search technique.Besides,two groups of experiments have been carried out to verify the result,the experiment of the nine digital finger language shows that the extracted features have rotation and scaling invariance.The experiment of twenty-three kinds of finger language shows that the proposed method has good Scalability.Experimental results show that the proposed method not only compensates for the shortcomings of the previous methods,but also retains the advantages of the previous methods.In this paper,the average recognition accuracy of the finger recognition method based on feature matching is 96.02%,which is invariant to rotation and contraction andinsensitive to illumination intensity.The improved finger language recognition method based on SVM not only has the above advantages,but also has good scalability,and the average recognition accuracy is 94.3%.