Corner Detection Research And In Medical Image Processing Applications

The image comer is an important feature of images, which contains massive informat-ion for the further processing. Comers can greatly reduce the computing complexity and, at the same time, keep some important features of an image. Therefore, comer det-ection has been a basic problem in image processing and plays an important role in m-any fields such as target recognition and feature matching.With the development of image processing technology, medical image processing con-stitutes a significant part in modern medical diagnosis and becomes an indispensable m-eans for the analysis. This paper combines medical images with corner detection metho-ds and gives a novel algorithm to detect corners in medical images, which will provide a theoretical basis for further medical image matching and feature extraction.This paper includes the following three aspects:1 First we will analyze the existed main corner detection algorithms. By reviewing t-he existed corner detection algorithms, the article makes a classification and analyze-s the advantages and disadvantages of these algorithms.2 Secondly, we generalize and classify the current application of corner detection algo r-ithms in medical image processing.3 Finally, based on SUSAN model, aiming at its defect that the model is dependent o-n the threshold greatly, this article puts forward a self-adaptive method to detect the candidate corners and a multi-level sift mechanism to decide the corners. By massive emulational experiments, the proposed self-adaptive sift algorithm is verified to be more efficient, more self-adaptive and more anti-noisy with lower time complexity, compared with Harris algorithm and SUSAN algorithm. The new algorithm is more applicable for complicated medical images.