Abstract
Purpose: According to Schmid's research[8], the Harris method[3] was still the best one available. We now present a method that, we believe, is somewhat better. In the full paper, we explain in much detail our method and the analysis of its effectiveness. In this abstract, we just add some pertinent remarks to listing the four topics of explanation. In the first topic, we analyze the intensity variations along pixel principal orientation axes and the analysis leads to classifying all pixels into four types in the full paper. The second topic is the flowchart of the algorithm of our method. Then we point out that our method is better than Harris method in three ways theoretically in the topic 3. At the last topic the experimental results and their analysis are given. The analysis shows that the locating error is much smaller than that of Harris method, as can be seen quantitatively from Table 1. Experimental results on real images show that our method can successfully detect almost all the corners of different kinds with good precision.
Original language | English |
---|---|
Pages (from-to) | 162-167 |
Number of pages | 6 |
Journal | Xibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University |
Volume | 26 |
Issue number | 2 |
State | Published - Apr 2008 |
Keywords
- Computer vision
- Feature extraction
- Intensity variation
- Principal orientation