Detection of Spot from Complex Regions Based on Geodesic Distance

Abstract

Spot detection is an important step in 2D-GE image analysis. The results of Spot detection may substantially influence subsequent stages of gel image analysis. However, in current suggested methods, the detection of spot in complex regions are far from satisfactory. This paper provides an effective method to raise the accuracy of spot detection in complex regions where many protein spots are overlapped. The proposed method firstly employs the Euclidean distance transform to estimate the spot centroids in complex regions and locates new centroids using the geodesic distance measure, then split the spot regions based on geodesic distances according the estimated spot centroids. compared with the existing method, experiments indicated the suggested method has good and efficient performance in the gel detection task.