Semi-Automatic Probabilistic Morphological Detection

We describe a semi-automated approach for designing morphological operators for detecting geometric structures in images. The approach automatically combines simple morphological primitives from a pool to generate an enumerable family of complex morphological operators that span a Receiver Operating Curve (ROC). Domain knowledge can be encoded by biasing the operator pool with primitive operators considered suitable by the designer; the system subsequently automatically selects and combines operators based on the joint statistics of the dataset. The major advantages of this method are that the designer can focus on constructing and analyzing individual generic simple operators, yet is able to rapidly combine them to yield more powerful, problem-specific solutions, whose operating point can easily be changed. We illustrate the approach by defining a generic probabilistic scene model using Birth and Death processes and associated simple operators, and combining the simple operators. Examples of using the system to perform video text detection are presented. Keywords— Morphology, automated design, preprocessing, probability, detection, bit-plane algorithms1.