Determining camera parameters from the perspective projection of a rectangle
Pattern Recognition
| Hi-index | 0.01 |
Abstract: ELISA plates are widely used to detect products of a proteic nature. Associated diagnostic tools are usually expensive; alternatively, modern pocket computers (such as PDAs and internet tablets) could become the platform of choice for low cost ELISA analysis software, whenever convenient user interfaces are developed. Election of a suitable input method is a key question in interface design for embedded devices. In this work, two input methods (on-screen finger and pen touching) for an application running on an internet tablet are compared. The user must click four times on a digital image represented on the screen, aiming at four calibrating points. The objective is to check whether one input method is significantly better than the other, according to their precision. A sample of digital images is available. An original mathematical procedure was developed which involves two steps for each input method and image: robustly fitting a rectangle to the four points, and calculating a special kind of distance between estimated and real points. Then, a statistical test for differences between both methods must be performed. The test chosen is based on a stochastic ordering. The results of this research lead to the conclusion that the pen input method is superior under the conditions of use of the application in question, so it is not advisable to suggest a finger-only user interface until further development is made on the display of on-screen images.