Research On Primary Corner Detection Algorithm

Extrac the feature information of image is one of the basis works in the image processing. Comapre with other image features which preserved the characteristics of the image,the corner point contain the smaller information data,less influence on the external factors, kept the feature information of the image with the least data. The study of the corner detection algorithm is therefore of great significance and have broad application prospects.One of the main work of this paper is the improvement of the Harris corner detection algorithm which is widely used at the present time. To solve the disadvantage of single scale and improve robust of the algorithm. The improvement approach of the Harris algorithm is based on Gaussian scale and gradient direction.the aim the improvement is to reducing the corner of the error rate effectively with less impact on the speed of the algorithm. Simulation results show that the improvement algorithm has good noise immunity in the noise affecting image,besides the corner detection is more stable and accurate positioning. Another focus of this research is Trajkovic corner detection,the result of this algorithm has a higher error proportion and to improvement the algorithm,this paper take the idea of local area to eliminate errors corners.The simulation results verified that the improvement of this paper have removed the situation of the hypotenuse abnormal response.