Hybrid of multi-signal processing methods for detection of small objects in containers filled with beverages using ultrasound.

摘要

Ultrasound backscattered signals for object detection could be too weak to be perceived when superposed to strong reflection signals, and could also be complicated either in time or frequency domain. These peculiarities raise a challenge for signal processing methods. In this paper four signal processing methods were studied to detect objects in containers. Root mean squares (RMS) method was better in object detection but needed to combine other methods to increase detection success rate. Variance, center frequency pressure ratio (CFPR) and backscattered amplitude integral (BAI) methods have similar overall detection ratings. The optimal signal size, to increase the detection ability, corresponded approximately to the length of reverberation signals before it was dissipated..