Multi-focal nematode image stack classification using a projection-based multi-linear method

Abstract

In digital multi-focal images (DMIs), morphological and topological information for a transparent specimen can be captured in the form of a stack of high-quality images. We propose to use projection methods such as coefficient of variation projection to exploit the entire information of a given DMI stack using its projection images from different directions. Besides, multiple features extracted from the projection images along different directions are combined by using canonical correlation analysis. Because DMI stacks represent the effect of different factors—texture, the directions of projection, different instances within the same class and different classes of objects, we embed the projection method within a multi-linear analysis framework to propose a multiple direction projection-based multi-linear classification approach. The experimental results on the nematode data show that our proposed classifier can achieve very reliable recognition rate (98.5%) on a real-life database, even we only use the texture feature instead of the combination of texture and shape features as in a previous work.

http://ift.tt/2z0HBJZ