In biological research images are extensively used to monitor growth, dynamics and changes in biological specimen, such as cells or plants. Many of these images are used solely for observation or are manually annotated by an expert. In this dissertation we discuss several methods to automate the annotating and analysis of bio-images. Two large clusters of methods have been investigated and developed. A first set of methods focuses on the automatic delineation of relevant objects in bio-images, such as individual cells in microscopic images. Since these methods should be useful for many different applications, e.g. to detect and delineate different objects (cells, plants, leafs, …) in different types of images (different types of microscopes, regular colour photographs, …), the methods should be easy to adjust. Therefore we developed a methodology relying on probability theory, where all required parameters can easily be estimated by a biologist, without requiring any knowledge on the techniques used in the actual software.A second cluster of investigated techniques focuses on the analysis of shapes. By defining new features that describe shapes, we are able to automatically classify shapes, retrieve similar shapes from a database and even analyse how an object deforms through time.