Acquisition Of 3D Foot Shapes Based On Stereo Vision

Acquisition of foot shapes has attracted extensive interests in recent years, which has potential application in a wide range of fields, such as medical diagnosis, rehabilitation exercise, fitness training etc. This thesis, based on the review to precious works, provides an integrated solution to acquire the foot shapes when it falls in static condition and walks in athletic process respectively according to the principle of stereo vision and digital geometry process. And then we design and develop a set of tools for customized shoemaking on the basis of research results and conclusions. The main contents focused by this thesis are as follows:1. A novel approach is proposed to scan 3D foot models by stereo vision techniques based on the explicit marker points woven in the socks.10 cameras are carefully configured to make sure that each marker point can be seen from at least two neighboring cameras. All marker points are extracted by the proper algorithm and 3D positions of all marker points are recovered in term of stereo vision principle. The 3D geometry of foot is acquired by the subdivision modeling based on the mesh composed of 3D marker points. The experimental results show that its precision is less than 1mm, which can well satisfy the requirement of customized shoe making.2. Starting from the request of customized shoemaking, presents an effective method for determining foot sizes. The method can obtain majority parameters of the foot sizes such as foot length, foot base width and the girth of different part. The error of this method is within 1mm.3. An active approach is presented to acquire 3D shapes of moving foot from multi-view video sequences based on the explicit marker points woven in socks. The reference 3D model of foot is firstly recovered from the starting frames in the multi-view video clips, and then the markers in each view are traced as a whole based on a continuous motion vector field built from the reference 3D model. The missing vertices in the candidate 3D model caused by self-occlusion, non-uniform illumination and random noises are reliably estimated and reconstructed in 3D space by minimizing the changes of differential features in 3D geometry of the reference model. The experimental results show that our method can robustly recover the 3D model of foot in complex background even if there is a relatively large movement between the adjacent frames, with an acceptable accuracy of the resulting 3D model.4. Based on the research achievement, a set of software for modeling and measuring 3D foot has been implemented and applied on practical production line. It serves for modeling foot, measuring foot, shoe-last design and also helps the management of customer information, e-Orders, operating records, and the information of foot models and shoe-last models, which greatly facilitates the entire production workflow of customized shoemaking.And the main contribution of this thesis includes:1. Unlike still objects such as sculpture, the foot is living organism and cannot remain perfectly still during the scanning. The mild spasm would interrupt the measurement and introduce significant error on the final result. The reported static foot shape acquisition approach can capture the foot in a split second, which can greatly reduce the error caused by the mild spasm and improve the result precision.2. The thesis solves the problem of acquiring the time varying foot shapes in the open environment from videos. While ensuring the result accuracy, the reported method can overcomes the difficulty of maintaining the final shapes completeness by exploiting the motion vector field to track the markers as a whole and estimating the missing marker points in 3D spaces based on the reference model, which also makes the tracking process independent of the background context around features and allows the algorithm to deal with the features with similar contexts in complex background. Furthermore, it can also cope with the problems under the conditions of sharp changes of the object shapes happened between consecutive frames even from badly polluted images. In most cases, our approach shows an obvious improvement over the existing methods that it can extract and track the markers followed by geometry reconstruction automatically except under some very extreme conditions.3. According to the shoemaking business processes, the thesis provides a complete solution of customized shoe making from foot acquisition to the output of digital shoe-last, which has been applied in actual production.