| Combining photoelectric technology to perform three-dimensional reconstruction of the foot,it can achieve the extraction of accurate foot size parameters for personalized footwear production;determine the arch type of the foot and use it for foot health diagnosis.Therefore,the design of the foot 3D reconstruction and measurement system is of great significance.This paper designs a set of foot 3D reconstruction and size measurement system based on Sf M and MVS.The system can realize the functions of scanning the foot with a single camera,reconstructing a complete foot model,extracting foot size parameter and judging the arch type.Foot scanning is the first and most important step to achieve 3D reconstruction.The scan result will directly determine the 3D reconstruction result.The goal of scanning is to obtain a complete model with the foot surface and sole,but a single parallel height surround scan cannot achieve this goal.This article designs three scan trajectories at different heights.One trajectory is used to scan the foot surface information,one trajectory is used to scan the plantar information,and the last trajectory is used to connect the foot surface and the plantar information.The image of the foot surface and the plantar can realize the feature matching connection through the middle layer image to realize the reconstruction of the complete point cloud model of the foot.Using images as input for point cloud reconstruction,the generation of background point clouds is inevitable.In order to reduce the interference of the background point cloud on the foot model,a mask device with low reflection intensity and easy installation is designed in this paper.The background of all images obtained after scanning is a non-reflective black mask.With the masked image as input,the number of background point clouds is greatly reduced,and the color information of the background point clouds is unified.Although the number of point clouds is greatly reduced after the masking process,the existing noise level still cannot meet the reconstruction conditions of a complete foot mesh model.This paper proposes a noise reduction method based on color clustering.The point cloud separation is achieved by using the difference in the spatial distribution of the foot point cloud and noise in the RGB three-dimensional space.Under the premise that the gray value of the noise is known,the first initial cluster center can be selected in the noise set by setting the gray value threshold and the fitting speed is accelerated;by setting an appropriate number of clusters,the foot point cloud and noise can be completely separated.This paper locates the characteristic points on the main axis of the foot model according to the law of foot shape and determines the foot length;Using the located feature points on the main axis and the cutting algorithm to extract the feature points on the foot surface,and determine the foot width,plantar girth,and hock girth;Using the cutting algorithm to intercept multiple cross-sections of the foot,and judge whether it is a high arch by judging the number of intercepted closed contours;Taking the longitudinal section of the first metatarsophalangeal toe,and judging whether it is a flat foot by calculating the distance between the highest point of the middle arc of the bottom contour line of the section and the lowest point of the foot. |
