Three-dimensional reconstruction of the cone-like aggregate pile based on binocular vision

Abstract

Abstract In order to use image recognition for the three-dimensional reconstruction of target objects faster and more conveniently, with the help of various machine vision technologies such as mathematical algorithms and machine learning algorithms, it aims to help engineers complete the three-dimensional reconstruction of cone-like aggregate piles. First, under the method based on a genetic algorithm, the proposed method can be used to identify the most common contour segments to deal with the contour recognition of aggregate piles and complete the work of region segmentation. The common fragment illustrates the particular logic contained in the outline. Then, this paper shows that the explicit representation of shape contour contributes to shaping representation and object recognition. Multiple two-dimensional virtual slices are used to divide the target object in the field of view of the binocular camera into multiple cross-sectional areas, so that an appropriate ellipse of the material pile contour curve similar to a cone is obtained to approximately express these cross-sectional areas. Finally, the three-dimensional reconstruction of the surface of the three-dimensional target object is completed by many two-dimensional elliptical slices. Experiments show that the method of three-dimensional reconstruction of images in a rapid and straightforward ability to prove its feasibility of the method.