Research On Mine Image Enhancement And Underground Personnel Detection

There are some dangerous regions in the complex environment of coal mine,the mistaken entry of personnel will cause harm to life.Considering the requirements for the safety production of coal mine,the automatic detection of personnel in underground coal mine videos to prevent dangerous behavior is of great significance.For lowillumination and uneven lighting of mine image,the methods of mine image enhancement and underground personnel detection are studied in this paper.The main research work and innovation includes:(1)A mine image enhancement algorithm based on homomorphic filtering and Curvelet transform is proposed to overcome the problem of low-illumination and uneven lighting.Homomorphic filtering is used to correct the image and adjust the brightness of the image;The original image and the corrected image are decomposed by Curvelet transform,and the high-frequency subband are fused by Sum-ModifiedLaplacian(SML)maximum method to ensure the clarity of image,and the low frequency subband are fused by region energy and variance product method to preserve more visual information of the image.The image restored by the inverse transform of Curvelet is processed with the region contrast enhancement method to improve the contrast of the image.The experimental results show that the method produces better result,the glaring regions are restrained and the dark regions are enhanced.(2)A multi-feature fusion miner detection algorithm based on adaptive local similarity model(ALSP)is proposed to overcome the problem of poor detection effect and high rate of missed detection and false detection.With the problem of noise sensitivity of traditional LBP,the method of LSP is improved by adaptive threshold to get better local texture.and GLCM features represent texture better.The ALSP features and GLCM features of image is fused to represent better image texture.Then,the HOG features are extracted and reduced by PCA.The final feature fused HOG features and detect the test image with the trained SVM classifier.The experimental results show that the method produces better detection effect in underground mine,and the rate of missed detection and false detection are reduced.