ABSTRACT
Features of an image are defined as regions where some unusual action happens. For example, the brightness changes drastically to form an edge; or, gradients of edges change drastically to form corner. This chapter explores how some of these features can be detected using convolution, which is a linear filtering process. Feature detection is considered to be part of low level processing in human vision. The chapter discusses the basic techniques to simulate human vision which have been combined to provide a much more sophisticated feature detector, relatively more popular of which is called the Scale Invariant Feature Transform. Probabilistic methods can also be applied to achieve better estimates by global relaxation labelling techniques. Feature edges are caused by the intensity moving from one level to another and then coming back close to the original. A general feature is a point of interest where neighboring patches overlapping the pixel have a high degree of variance.
