New vision location algorithm based on one straight line and two points features

Generally, we use point, line and conic features to locate object. At present, most pose estimation method use one kind of feature to locate object, but not combine different features to locate. In some circumstances, the numbers of one kind of feature maybe not enough to locate, but there exit different features in natural surroundings. In this case, using different features to locate is convenient. Thus, we propose this new method which combines different features of one line and two points to locate object. The numbers of feature to locate is three, which is the minimum configuration. Two random points can form one line. When the line formed by two points is parallel to the other line, we can compute the rotation and translation matrix. This method we propose can obtain the closed-form solution of 3D pose and it is simple. Therefore, it has good real-time characteristic in locating.