| Summary: | 碩士 === 國立成功大學 === 數學系應用數學碩博士班 === 106 === In recent years, spatial positioning has been widely used in various fields.
This study intends to spatially locate images captured by a digital camera to meet the needs of popularization. The main research object of this paper is that when the size of the target is known, we would like to locate the target through a single image captured by a digital camera. Through the size of the object and the position in the image of the object, different object distances can be calculated for different object plane normal vectors. If the object plane equation is ax + by + z = z0, then let the nonlinear function contains three parameters, respectively a; b; z0, which is the total sum of squares of the object distances. When the correct parameters are brought in, the function value will be 0. In order to solve the nonlinear function, the camera’s internal parameters must be calculated with other image for correction before capturing the image for positioning. The numerical methods used are gradient method and Newton method, the accuracy between the two algorithms and the number of iterations required are compared under different conditions.
According to the experimental results, the gradient descent method with better convergence stability will be the main iterative method.
|
|---|
