Abstracto

An improved Automatic Virus particle Detection method based on adaptive K-NN classifier

Mr. Dinesh Kumar M.E., Desmond John

Several automatic and semi-automatic particle detection algorithms have been developed along the years. Here we present a general technique designed to automatically identify the projection images of particles New methods are described that should facilitate high-resolution (5–10 Å) image reconstructions from low-dose, low-contrast electron micrographs and processing of large, digital images produced by new imaging devices and modern electron microscopes. Existing techniques for automatic selection of images of individual biological macromolecules from electron micrographs are inefficient or unreliable. The initial detection of the particles takes place through automatic segmentation of the entropyproportion image; this image is computed in particular regions of interest defined by two concentric structuring elements contained in a small overlapping window running over the entire image. Morphological features help to select the candidates, as the threshold is kept low enough to avoid false negatives. The candidate points are subject to a credibility test based on features extracted from eight radial intensity profiles in each point from a texture image and Gabor image. For candidate selection use high level Adaptive K-Nearest neighbor Classifier. It’s improving the accuracy of candidate selection with the need of low level set features.