Multi-surface Pattern Separation of Interdigital Dielectrometry Data for Volume Detection

The use of multi-wavelength interdigital sensors for non-destructive testing is based on the capability of the measuring system to classify the measured impendence according to some physical properties of the material under test. This is done mostly with computational techniques since the inverse problem is ill-posed and sometimes leads to physical unaccepted solutions. Thus, the successful application of interdigital sensors are related to the construction of a set of measurements sufficient to classify the physical property of interest. A separable measuring data set consists of measurements obtained in a predefined set of M-sensor wavelengths (multiples of a base wavelength) and in a predefined set of N-frequencies. The classification of the MXN-dimensional vectors obtained in this way can be used to detect volumes of different shapes and sizes inside a uniform medium. The multi-surface pattern separation techniques has been applied, showing that the detection of a volume is possible if the radius of the volume (considering that the volume is spherical) is comparable with the base wavelength of the sensor device.