Research On Key Technology Of Three Dimensional Photon Counting Integrated Imaging

Three-dimensional photon counting integrated imaging can realize 3D reconstruction of target scenes in low-light environments,and is widely used in the fields of national defense,military and national economic construction.In this paper,the research on photon counting integrated imaging is carried out,focusing on key technologies such as image preprocessing,image matching and surface reconstruction in the reconstruction process.In order to realize the 3D reconstruction of photon counting integrated images,the Multi-pixel Photon Counter(MPPC)is used to build an experimental platform for photon counting integrated imaging according to the principle of off-axis distribution-aware integrated imaging,and to realize the acquisition of 3D target object images in a low-illumination environment.Aiming at the problems of low brightness of images captured by the system in low-illumination environment and also accompanied by severe noise,which affects the subsequent 3D reconstruction effect,an edge detection algorithm for low-illumination images based on the parametric logarithmic image processing model is proposed,which uses the Parametric Logarithmic Image Processing(PLIP)model theory to derive a new gradient operator,replaces the derived new gradient with the gradient in the traditional Canny algorithm,and applies the three best edge detection criteria of the Canny operator to extract the edges of the images.The experimental results show that the edge detection effect of the new algorithm is significantly improved compared with other algorithms,and the edge information of the image detected by the algorithm is used to enhance the edge of the collected image,which effectively improves the clarity of the image.In the process of 3D reconstruction,matching algorithm is prone to redundancy,uneven distribution and poor self-adaptability in extracting low-contrast image feature points,etc.An improved ORB(Oriented FAST and Rotated BRIEF)algorithm with high degree of self-adaptability and uniform feature distribution is proposed for extracting image feature points.algorithm.The algorithm calculates the number of layers to build the required image pyramid based on the image size;adaptive region division is performed for each pyramid image layer,and the adaptive extraction threshold of FAST corner points is set by using the image contrast of each region;finally,the improved quadtree algorithm is used to eliminate the redundant feature points and even out the distribution of feature points.The experimental results show that the improved algorithm effectively improves the distribution uniformity,extraction efficiency and matching accuracy of feature points.The extracted feature points are used for 3D coordinate calculation to obtain the point cloud image of the target object.By comparing the generated point cloud images,it is verified that the image preprocessing technology and the improved feature point matching algorithm in this paper improve the reconstruction effect of the 3D point cloud.Since the discrete point clouds cannot show the details of the object surface completely and truly,an optimized Delaunay triangular dissection algorithm is designed to dissect these discrete point cloud data,build a mesh skeleton model,and then take out the textures of the object from the image and map them to the skeleton model to finally realize the 3D reconstruction of the image.The effectiveness of the algorithm proposed in this paper is further demonstrated by experimental comparative analysis.