Horizontal Vanishing Point Research Articles

3D vehicle model-based PTZ camera auto-calibration for smart global village

PTZ cameras perform a critical role in highway surveillance systems because of their wide field of vision and high degree of flexibility. However, it is difficult to obtain accurate real-world physical information by using the acquired image information due to changes in focal length and angles of the PTZ cameras. Therefore, it is of great value to investigate off-site auto-calibration methods for PTZ cameras in highway surveillance system applications. This paper proposes a PTZ camera auto-calibration method based on vanishing points and 3D vehicle model constraints to establish an accurate description of the relationship between image information and real-world physical information. Firstly, the position of the longitudinal vanishing point is estimated using the cascaded Hough transform voting of the vehicle target motion trajectories. Secondly, the physical size of the 3D vehicle model is used as a constraint, and it is combined with an enumeration strategy to obtain accurate estimation of the horizontal vanishing point. Finally, the calibration parameters of the PTZ camera are calculated with the known camera height. Experiments on numerous sets of highway surveillance scenes demonstrate that the proposed PTZ camera auto-calibration method is able to satisfy the application requirements of highway surveillance scenes. Compared with reference methods, this method exhibits superior performance. Moreover, this method possesses substantial value for various applications.

Vanishing Point Detection in Urban Scenes Using Point Alignments

We present a method for the automatic detection of vanishing points in urban scenes based on finding point alignments in a dual space, where converging lines in the image are mapped to aligned points. To compute this mapping the recently introduced PClines transformation is used. A robust point alignment detector is run to detect clusters of aligned points in the dual space. Finally, a post-processing step discriminates relevant from spurious vanishing point detections with two options: using a simple hypothesis of three orthogonal vanishing points (Manhattan-world) or the hypothesis that one vertical and multiple horizontal vanishing points exist. Qualitative and quantitative experimental results are shown. On two public standard datasets, the method achieves state-of-the-art performances. Finally, an optional procedure for accelerating the method is presented.

소실점 추정 정확도 개선을 통한 SVM 성능 향상

In this paper, we propose an improved single view metrology (SVM) algorithm to accurately measure the height of objects. In order to accurately measure the size of objects, vanishing points have to be correctly estimated. There are two methods to estimate vanishing points. First, the user has to choose some horizontal and vertical lines in real world. Then, the user finds the cross points of the lines. Second, the user can obtain the vanishing points by using software algorithm such as [6-9]. In the former method, the user has to choose the lines manually to obtain accurate vanishing points. On the other hand, the latter method uses software algorithm to automatically obtain vanishing points. In this paper, we apply image resizing and edge sharpening as a pre-processing to the algorithm in order to improve performance. The estimated vanishing points algorithm create four vanishing point candidates: two points are horizontal candidates and the other two points are vertical candidates. However, a common image has two horizontal vanishing points and one vertical vanishing point. Thus, we eliminate a vertical vanishing point candidate by analyzing the histogram of angle distribution of vanishing point candidates. Experimental results show that the proposed algorithm outperforms conventional methods, [6] and [7]. In addition, the algorithm obtains similar performance with manual method with less than 5% of the measurement error.

Using pedestrians walking on uneven terrains for camera calibration

A calibrated camera is essential for computer vision systems: the prime reason being that such a camera acts as an angle measuring device. Once the camera is calibrated, applications like three-dimensional reconstruction or metrology or other applications requiring real world information from the video sequences can be envisioned. Motivated by this, we address the problem of calibrating multiple cameras, with an overlapping field of view observing pedestrians in a scene walking on an uneven terrain. This problem of calibration on an uneven terrain has so far not been addressed in the vision community. We automatically estimate vertical and horizontal vanishing points by observing pedestrians in each camera and use the corresponding vanishing points to estimate the infinite homography existing between the different cameras. This homography provides constraints on intrinsic (or interior) camera parameters while also enabling us to estimate the extrinsic (or exterior) camera parameters. We test the proposed method on real as well as synthetic data, in addition to motion capture dataset and compare our results with the state of the art.

Rectifying perspective views of text in 3D scenes using vanishing points

Documents may be captured at any orientation when viewed with a hand-held camera. Here, a method of recovering fronto-parallel views of perspectively skewed text documents in single images is presented, useful for ‘point-and-click’ scanning or when generally seeking regions of text in a scene. We introduce a novel extension to the commonly used 2D projection profiles in document recognition to locate the horizontal vanishing point of the text plane. Following further analysis, we segment the lines of text to determine the style of justification of the paragraphs. The change in line spacings exhibited due to perspective is then used to locate the document's vertical vanishing point. No knowledge of the camera focal length is assumed. Using the vanishing points, a fronto-parallel view is recovered which is then suitable for OCR or other high-level recognition. We provide results demonstrating the algorithm's performance on documents over a wide range of orientations.