Research On Feature Extraction Method Of Dangerous Rock Cracks Fracture Based On Uav Image

China is one of the countries most seriously affected by landslides in the world.Most landslides occur in dangerous rock areas.Cracks in dangerous rocks are the main diseases of dangerous rocks.High-resolution digital images of dangerous rock surface can be obtained by UAV aerial measurement technology,crack detection on images can provide index information for dangerous rock monitoring.This paper mainly studies the method of extracting the characteristics of dangerous rock fractures from the digital image of Uav dangerous rock mass.The main contents are as follows:First,a method for extracting crack feature based on Butterworth high-pass filter and histogram matching is proposed.Firstly,the original gray image is filtered by median filtering to remove the discrete noise in the image;Then using Butterworth high-pass filter in the frequency domain filter out low frequency information,enhance fracture edges;Next,empirical threshold segmentation is carried out to obtain the binary image,and the binary result and the original gray image are matched to extract the crack information through histogram matching,but there is still a small amount of noise.;Finally,exponential transformation is used to further remove the noise,and the ideal result of dangerous rock fracture extraction is obtained.Second,a crack feature extraction method based on mathematical morphology and Prewitt operator is proposed.Firstly,the edge of crack is enhanced by close and open operations of gray morphology;Then median filtering is performed to remove the discrete noise;Then the multi-directional Prewitt operator is used to extract the crack;And then the histogram threshold segmentation is carried out to extract the edge information of the crack,but there is still a small amount of noise in the figure;At last,the open and close operations of binary morphology are used to de-noising,and the ideal results of dangerous rock fractures are obtained.Combined with the advantages of the horizontal and vertical Prewitt operators and the diagonal Prewitt operators,the multi-directional enhancement effect of the information of dangerous rock fractures can be realized.Third,According to the two methods mentioned above,an experimental system for feature extraction of dangerous rock mass fracture was developed by using graphical user interface GUI in Matlab software.The system includes four modules: document,file image preprocessing,crack feature extraction method based on high-pass filter and histogram matching,and crack feature extraction method based on mathematical morphology and Prewitt operator.It provides common methods in image preprocessing,and shows the effect of the above two methods in stages.The experimental results show that both methods can obtain relatively clear fracture features,and the experimental system is characterized by simple operation,fast speed and friendly interface,so the extraction results can be obtained quickly.Will the method applied in Chongqing the nanchuan ZengZi YanWei rock zone in information collection in dangerous rock crack has obtained the good effect,has a certain application prospect in the field of monitoring dangerous rock.