Reconstruction And Measurement Of Fabric Drape Based On Image Sequences

Fabric drape refers to the phenomenon that a fabric will become deformed when it is subjected to gravity. It is one of the most important indicators which influences the comfort and aesthetic appearance of the fabric. Until now, there are many investigations about fabric drape at home and aboard, the indicators using in these studies usually depend on the configuration of the draping fabrics. As an important factor of the draping fabrics, the three dimensional shape of the fabric should be taken into account in order to evaluate the drapability of a fabric comprehensively. However, only a few literatures have focused on this work and most of their reconstruction processes are complex and inaccurate.In this paper, a simple and useful method is proposed to reconstruct 3D colored models of draping fabrics using image sequences captured by the camera of a smart phone. Moreover, the drape parameters can be extracted according to the projection of the point cloud of the drape model. In our method, SIFT(scale-invariant feature transformation) and Harris corner detection methods were adopted to extract the feature points from the images. These points will be used to generate consistent tracks which were further processed to generate3 D point cloud of the draping fabric.The work have been done in this research mainly includes:1. This paper combined the reconstruction of fabric drape and the theory of computer vision together well. The image reconstruction technique has many advantages such as low price, simple operation, better reconstruction. It is a good choice for fabric drape reconstruction. By studying and analyzing the shape of draping fabrics and the principle of reconstruction based on image sequences, an appropriate trace was designed for image acquisition to reconstruct a better model.2. A drapeometer with checkerboard was made to measure the 3D configuration automatically. The checkerboard provides a base plane and scale for the reconstructed model, which can help to rotate, translate or zoom the fabric drape model in the world coordinate system accurately. This accomplished the extraction and calculation of the drape parameters automatically in the later steps.3. A new method was proposed to calculate fabric drape parameters according to the 3D fabric drape model. Generally, the traditional methods extract the parameters from the orthographic projection image of the draping fabric acquired by an CCD camera. However, the results should be revised due to the phenomenon of perspective. Therefore, it is easy to have a big error when the correction factor is not suitable. In our method, the drape projection is computed in 3D coordinates, which would be more reliable compared to the projection obtained by the CCD camera.4. To verify the feasibility of the proposed method, the correlation analysis and regression analysis were applied to compare the measurements based on XDP-1 drapeometer and the measurements based on the 3D drape model. The results indicated that the drape parameters extracted from our method is reliable.