Papers by Keyword: Image Matching

Paper TitlePage

Authors: Zhong Yan Liu, Guo Quan Wang, Dong Ping Wang
Abstract: A method was proposed to gain three-dimensional (3D) reconstruction based on binocular view geometry. Images used to calibrate cameras and reconstruct car’s rearview mirror by image acquisition system, by calibration image, a camera's intrinsic and extrinsic parameters, projective and fundamental matrixes were drawn by Matlab7.1;the collected rearview mirror images is pretreated to draw refined laser, extracted feature points, find the very appropriate match points by epipolar geometry principle; according to the camera imaging model to calculate the coordinates of space points, display point cloud, fitting space points to reconstruct car’s rearview mirror; experimental results show this method can better restore the car’s rearview mirror of 3D information.
Authors: Feng Chen Qian, Ning Shan, Ya Lin Ye, Yong Zhong Ma, Xue Yan Hu
Abstract: The paper proposed a data compression method in multiresolution structure and realized multiresolution rendering of point model by traversing the KD-tree breadth-first. It established the details control of multiresolution rendering, improved the drawing efficiency of point model. And a simulation study was carried out. Results show that the method could not only reduce the geometry data volume of point models directly, but also reduce the sampling point number without debasing the numerical precision of the model geometry data, so as to further reduce data volume of point model. In addition, the mixed rendering of geometric model point and rectangle in this method overcome the shortcoming that point model set is unsuitable for large flat surface.
Authors: Yan Lu Xu, Yan Ma, Shun Bao Li, Ning Li Zhang, Xiang Fen Zhang, Yu Ping Zhang
Abstract: This paper presents a new algorithm of image matching via combining sift and shape context for improving image matching accuracy. A joint descriptor is applied to describe feature points. The initial matching is obtained by a proposed distance formula. Furthermore, PLS is introduced to eliminate mismatched points. Experimental results demonstrate the proposed algorithm can achieve better performance compared to conventional methods.
Authors: Qian Zhang, Ning Shan, Feng Chen Qian, Ya Lin Ye
Abstract: According to the space equal features of midpoint segmentation KD-tree, anisotropic quantitative method is proposed. After having used this method, the coordinate figures of sampling points in point model were quantified again; and the quantitative results have been represented as encoding of space partition methods and results in the process of constructing KD-tree. Meantime, the effectiveness that the method was used to express point model have been simulated and studied. It turned out that the quantitative method is very suitable for large flat area, because it is not only directly reduce the amount of point model geometry data, but also greatly reduce sampling points in case of keeping the numerical precision of the model geometry data unchanged, so as to further reduce the amount of point model data.
Authors: Cheng Zhi Cao, Wen Zheng Gao, Hui Jun Li
Abstract: In a vision-based system for mobile robot/human localization in the environment, extracting feature from images and image matching are important for the final results. Although there are many algorithms for image matching, they are not fit for this system. A new method is designed to improve image matching effect.
Authors: Guang Sheng Liu, Ben Jun Zhang, Meng Yang
Abstract: The measurement of muzzle vibration responses during gun firing is one of the most important factors for successful design of gun. The use and result are limited by traditional method for the effect of firing environment, so new measurement method is required, the high-speed imaging technology is one choice. The application of high-speed imaging technology in measurement of muzzle vibration is focused on qualitative analysis, in order to discuss the quantitative analysis in this paper, images of muzzle vibration responses are obtained by using high-speed vidicon. Image pretreatment is done for post processing expediently, the process of pretreatment include de-noising by using wavelet transform method, gray processing by using averaged-value method and contrast improvement by using wavelet transform method. After image pretreatment, orientation and matching is done by using correlative matching method, and then the numerical of muzzle vibration responses during gun firing are obtained. So a novel method is afforded for measuring of muzzle vibration responses during gun firing, and the new method has practical significance in the gun’s design, research and production.
Authors: Jing Hou, Jin Xiang Pian, Ying Zhang, Ming Yue Wang
Abstract: A new approach is presented to match two images in presenting large scale changes. The novelty of our algorithm is a hierarchical matching strategy for global region features and local descriptors, which combines the descriptive power of global features and the discriminative power of local descriptors. To predict the likely location and scale of an object, global features extracted from the segmentation regions is used in the first stage for an efficient region matching. This initial matching can be ambiguous due to the instability and unreliability of global region feature, and therefore in the later stage local descriptors are matched within each region pair to discard false positives and the final matches are filtered by RANSAC. Experiments show the effectiveness and superiority of the proposed method in comparing to other approaches.
Authors: Xing Xiong, Byung Jae Choi
Abstract: SURF (Speeded Up Robust Features) is known to be a famous and strong but computationally still expensive.It has not attained real-time performance yet. In this paper we analysis the SURF in orientation and descriptors extraction method forresolvingsome problems. For example, matching images through the SURF algorithm spends too much time and causes some errors by integral images. We propose a novel orientation and descriptor algorithm to improve the conventional SURF. Theproposed method shows some advantages such as a faster speed.
Authors: Xiu Yuan Xing, Dai Zhuang Wang, Peng Yi
Abstract: In this paper, the binocular stereo system is applied to intelligent vehicle. A series of simulation tests have been carried out. Two CCD video cameras with the consistency of parameters and performance are applied in the tests to acquire image synchronously. In order to achieve the performance of real-time and accuracy, Hough Transform is applied for edge detecting and SURF algorithm is used for image matching. After image binarization, edge detection and image matching, we can calculate the parallax of the two corresponding images, and then get the distance and relative velocity of the obstacle. Satisfactory results have been achieved in comparison with the reality.
Authors: Yi Cui, Xian Zhao, Chang Feng Jing
Abstract: . This paper introduces a new approach on measurement of aerial photogrammetry. Because of the development of stereoscopic display technology, the author wants to change 2D working environment to 3D working environment during aerial photogrammetry measuring process. The approach gives a new way to do aerial photogrammetry measurement in 3D environment. Firstly, we made image pixel data one-to-one corresponding with DEM data of the area. Then, we built a 3D model of the area. Since the one-to-one corresponding of DEM and image, we obtained photo coordinates of the target points of stereo pair on left image and right image in single click. And then, with the help of traditional theory, we finished calculation of the transforming from photo coordinate system to ground coordinate system, by which we could obtain ground coordinates of the target points. Finally, we overlapped the points which we calculated on the DLG in ArcGIS software to validate the correction of the calculation. The result shows that the precise of this approach is acceptable. At the end of the paper, the author gave some advices to improve the approach and the expectation of 3D measurement in true 3D environment.
Showing 1 to 10 of 76 Paper Titles