Ning Wei

Simple 3-D Point Reconstruction Methods With Accuracy Prediction for Multiocular System

For the demands of surgical navigation, this paper proposes some simple 3-D point reconstruction methods. These methods regard the perpendicular feet on back-projection lines as the measurements of a 3-D point, since these feet are close to the 3-D point. And then the methods utilize the distance between 3-D point and camera and error propagation rules to adjust the weights of the feet. Finally, the weighted average of these feet is taken as the final estimation for the 3-D point. They are easy to implement, and can be used in both biocular systems and multiocular systems. Especially, as these methods have closed-form solutions, their errors can be predicted by using error propagation rules. Experiments show that they are faster and more accurate than iterative methods and their error covariance matrix can be exactly predicted.

Enhanced Epipolar Constraint for Point Correspondence in Computer Vision System

Some projective points (2D-points) of the same 3D-point captured by different cameras are called corresponding points. The task of finding the corresponding points is named as point correspondence. It is a essential problem in computer vision, and it paves the way for 3D reconstruction. Epipolar constraint is a widely-used geometric constraint for point correspondence, some methods based on it have been proposed for this task. However, the threshold is set by empirical approach in these methods; this approach is influenced by the parameters of cameras. On the other hand, epipolar constraint is not a sufficient condition for this task; this will result in the mismatch for two 2D-points. According to the rule of Propagation Uncertainty and practical situation of the computer vision system, this paper introduces the enhanced epipolar constraint comprising normalized coplanar constraint, nonparallel constraint and interval constraint. The normalized coplanar constraint provides a simple and rational method to set the threshold of epipolar constraint; the nonparallel constraint and interval constraint can reduce the occasion of mismatch.

Angle Constraints for Point Correspondence in Multi-Ocular Vision

Geometric constraints have an essential significance for point correspondence in computer vision systems. Traditional epipolar constraint in bi-ocular system faces two main problems: threshold-setting and corresponding ambiguities. This paper describes a collinear epipolar plane model and proposes a novel criterion for bi-ocular corresponding which allows setting a uniform threshold. Furthermore, it proposes the concept tri-correspondence units, proves their specificity against ambiguities, and discusses the merging of them.

The design of infrared touch screen based on MCU

This paper introduces a touch screen technology based on infrared optical. This technology has apparent advantages in the large-size applications, which is simple, low cost, high feasibility. In the system, the detection of ambient light and the adaptive control of IR LED can effectively enhance the adaptability of the touch screen in complex light environment via the MCU. In image processing, the image of foreground and background are segmented by Otsus method with a coefficient. Integral projection is adopted to recognize the contact area. The result shows that the performance of the system fully meets the requirements of real time and accuracy.

A novel correspondence searching strategy in multiocular vision

Correspondence searching among different images is a fundamental problem in computer vision. It is important to find correspondences correctly and rapidly, especially for real-time tracking systems. Therefore, the definition of search areas in images is crucial. Traditional epipolar constraint is not noise-enduring; some reformative methods lack explicit geometric meanings. All of them cannot help defining rational search areas under noises. This paper proposes two new binocular imaging constraints with clear geometric meanings and strong restraining forces. Based on them, a novel searching strategy among multiimages is developed which can define optimal search areas with smallest sizes but best reliability. Practical algorithms for implementation are presented and experiments with real images are performed, validating the effectiveness of the proposed strategy.

A new real-time method for distortion correction in surgical robot positioning systems

A surgical robot usually utilizes a small marker ball to implement positioning task. However, a tough problem of this system is that images gathered in this system are always distorted, which reduces the accuracy of our positioning. Conventional distortion correction methods have to compute the whole image before determine a certain points ideal position, which is time consuming and cannot meet the systems real-time requirement. So in this paper, a novel correction method is proposed. It is distinguished from the conventional methods in its real-time property. This novel method analyzes the distortion model, to establish an equation between the distorted points and the distortion-free points. Then an algorithm based on fixed point theorem is applied to solve this high order equation. Experiments with explicit data show that this novel method has about the same correction accuracy as the conventional methods but with much less time, totally meet the real-time requirement.