REAL-TIME IMAGE-PROCESSING BASED ON NONLINEAR MODELS WITH APPLICATIONS TO IMAGE-ENHANCEMENT, AREA DETECTION AND VOLUME VISUALIZATION

This paper presents non-linear filter models that are suitable for real-time operations. The requirements imposed on these models by ultrasonic imaging are discussed. In particular, we study the characteristics of temporal filtering in cardiac imaging. Non-linear filter models incorporating these requirements are presented and compared with filtering schemes traditional in ultrasonic imaging. Furthermore, we illustrate how these filter models can be utilized to improve the quality and noise tolerance of automatic, real-time area detection algorithms. Finally, we present results from a 3D ultrasonic study of a foetus. The experiments illustrate how non-linear processing can increase the clinical information content that can be extracted from such investigations.