Research On Critical Technology Of Intensity Imaging Of 905nm Pulsed LiDAR

With an ever growing of LiDAR,the classification and recognition of target based on LiDAR have attract various attention in the literatures.Recently,it is extremely difficult to classify the targets at the same distance with the three-dimensional distance information the LiDAR provided.As we all know,the echo intensity value of the LiDAR represents the reflection characteristics of the target,and the echo intensity value is not affected by the dark condition,which made up the disadvantages of threedimensional distance image and the visible light image.Therefore,the echo intensity is applied into the various fields of classification and pattern recognition of target.According to the theory of target reflection characteristics and laser transmission characteristics,the key technologies of the 905 nm pulsed LiDAR are studied in this paper.Firstly,the theoretical derivation model of the LiDAR echo power for the point target and the extended target is established,which lays the foundation of the LiDAR intensity images.Secondly,the influence of rough surface on the laser echo intensity distribution is analyzed with the Monte Carlo method and random rough surface model.Moreover,A LiDAR echo intensity test platform is built to analyze the influence of distance,color,area and roughness on echo intensity.Next,the integral centroid method,the centroid-based intensity mapping normalization algorithm and the threshold-based K-means clustering algorithm are proposed respectively for the extraction,processing and classification of LiDAR intensity values.The experimental results demonstrate that the integral centroid method scales well the extraction of laser radar intensity compared with the peak method,and the accuracy is improved by 33%.Besides,the centroidbased intensity mapping normalization algorithm achieves better normalization of intensity values for different distances and angles,while the threshold-based K-means clustering algorithm is more reasonable and accurate than the traditional K-means clustering algorithm.Furthermore,the point scanning platform is designed and the intensity imaging and target classification are obtained by the above three algorithms.Finally,the 8*8 APD array LiDAR imaging platform is set up to acquire the LiDAR intensity imaging,and the threshold-based K-means clustering algorithm is utilized to acquire the target classification according to the distance information and the intensity information.In this paper,the proposed approach acquires the intensity values of 905 nm LiDAR effectively and provides an accurately classification results of the specific target.In summary,this research has practical significance for the application of LiDAR imaging system and intensity imaging.