Millimetre-wave radar in a missile seeker for the engagement of ground targets allows all-weather, day and night, surface imaging and has the ability to detect, classify and geolocate objects at long ranges. The use of a seeker allows intelligent target selection and removes inaccuracies in the target position. The selection of the correct target against a cluttered background in radar imagery is a challenging problem, which is further constrained by the seeker’s hardware and flight-path. This thesis examines how to make better use of the components of radar imagery that support target selection. Image formation for a squinted radar seeker is described, followed by an approach to automatic target recognition. Size and shape information is considered using a model-matching approach that is not reliant on extensive databases of templates, but a limited set of shape-only templates to reject clutter objects. The effects of radar sensitivity on size measurements are then explored to understand seeker operation in poor weather. Size measures cannot easily be used for moving targets, where the target signature is distorted and displaced. The ability to detect, segment and measure vehicle dimensions and velocity from the shadows of moving targets is tested using real and simulated data. The choice of polarisation can affect the quality of measurements and the ability to reject clutter. Data from three different radars is examined to help to understand the performance using linear and circular polarisations. For sensors operating at shorter ranges, the application of elevation monopulse to include target height as a discriminant is tested, showing good potential on simulated data. The combination of these studies offers an insight into the performance factors that influence the design and processing of a radar seeker. The use of shadow imagery on short-dwell radar seeker imagery is an area offering particular promise.