Classification of ultrasonic indications using pattern recognition

A method for identifying characteristics of a target, the method comprising:

• providing an analyzer configured to connect with one or more ultrasonic sensors;
• receiving, at the analyzer from a memory, data including ultrasonic measurements of a target, the received ultrasonic measurements being acquired by the one or more ultrasonic sensors and representing an incident ultrasonic signal reflected from a target;
• receiving, at the analyzer from the memory, a first classifier configured to recognize an ultrasonic pattern representing a target characteristic from the received ultrasonic measurements;
• comparing, using the analyzer and the received classifier, at least a portion of the received ultrasonic measurements to the one or more ultrasonic patterns of the received first classifier;
• recognizing, by the analyzer, ultrasonic patterns contained within the received ultrasonic measurements that substantially match the one or more ultrasonic patterns of the received first classifier to identify the target characteristic;
• and
• providing, by the analyzer, the received ultrasonic measurements and the identified target characteristic to a trainer;
• receiving, by the trainer from a database, one or more second classifiers that are different from the first classifier, wherein each of the second classifiers is configured generate an estimated identification of the target characteristic based upon recognition of the one or more ultrasonic patterns from the received ultrasonic measurements;
• receiving, by the trainer, a known identification of the target characteristic;
• determining by the trainer for each of the second classifiers, an error e1 between the estimated identification of the target characteristic and the known identification of the target characteristic;
• determining, by the trainer, an error e2 between the identification of the target characteristic for the first classifier and the known identification of the target characteristic;
• and
• updating, by the trainer, the first classifier with the second classifier having the error e1 when the error e1 is less than the error e2.