Study On The Slope Monocular Recognition Method And Algorithm Based On The Displacement Field

Landslide caused by slope instability is a common geological catastrophe in our country.Landslide in the event,often leads to significant loss of life and property.To study the convenient and practical dynamic detection method of slope,which is the key and premise to reducing landslide disaster successfully.The existing detection methods,such as the GPS method,laser 3D imaging method,artificial field method,sensor method,are equipment expensive,high degree of specialization,high cost of detection,big workload in the industry in different extent.In this article,we take the slope monocular recognition method based on the displacement of the slope to test the displacement,to find a new method for detection by making full use of the advantages of single camera imaging and digital image analysis technology.Firstly,it is a detection method of non-contact and without being limited by the range,which will not hurt the measured object and can reduce the risk of staff at the same time.Secondly,it has the advantages of collecting data fast,high precision,large information capacity,reliable and reusable data storage.Again,it has the ability of multipoint measurement at the same time,which can fully reflect the slope deformation dynamically.Finally,monocular detection is low-cost,convenient in project application and don't need later maintenance.Monocular recognition method based on displacement matrix of slope,which reconstructs the spatial coordinates of the target by the camera imaging information of a single camera.To achieve such a goal,we need to set the artificial markers on the surface of the slope to build a slope displacement field,use the image processing and identification technology for parameter identifications of markers in the displacement field.The core of the method is to identify the spatial coordinates of the marker array,which generally recognize the markers at the first place to achieve the markers' plane projection parameters,and calculate the corresponding spatial parameters with the knowledge of the three dimensional reconstruction method,to gain the spatial data of the markers array.Based on the research train above,we do our research work on the following aspects in this article.First,the design of slope displacement field based on the detection of marker' characteristics.The design of slope displacement field includes the design of the markers' size,shape,color,material,the design of the layout form of the markers in the slope displacement field,the rational selection of the observation field and the the establishment of coordinate system.In this article we set M × N red markers on the slope surface in the form of M × N array,which are called observation markers,to lay out a displacement field.In addition,we set two other red markers of localization in the stable locations of the slope surface,which are used for the built of coordinate system automatically and calculating model.Second,the valid recognition of markers based on the color and shape of markers and the detection of the markers' characteristic parameters.In the terms of the valid recognition of markers,we build the two-step recognition model based on the color and shape of markers.First of all,we make full use of the transformation relationship between the RGB color space and HSI color space and the red's distribution rule of HSI color model to achieve the initial recognition of the markers in H domain.Then,filtering the resulting image with morphological filters to reduce the influence of such factors as illumination,using Canny operator for edge detection.Finally,mix round/ellipse detection algorithm is employed to recognize the shape of markers to achieve the secondary recognition.In terms of the detection of the markers' characteristic parameters,the internal rotation scanning method is developed to detect the characteristic parameters(area and barycentric coordinates).The external spiral scanning method is developed to detect the characteristic parameters when the markers' rotating or rupture occur at the same time.Third,the design of the algorithm of 3D parameters calculation.Based on the pinhole model and the relationships between positioning markers and observation markers,we work out a conclusion that the square root of the object image area is inversely proportional to the depth value and calculate the depth of each marker with the known parameters of positioning markers.Then,we calculate the distances between the observation markers and positioning markers by the principle that the image pixel displacement between two points always keep proportional relationship corresponding to the actual displacement of the two points.Finally,the 3D coordinates are calculated by using the three-point calibration principle.