A Novel Change Detection Technique for Occluded Objects using Ultra-Wideband Radar
Soumya Chakravarty, Smriti Rani, Arijit Chowdhury, Tapas Chakravarty, Arpan Pal · 2021 29th European Signal Processing Conference (EUSIPCO) · 2021
The burgeoning e-commerce business model has witnessed several cases where the packages with faulty goods, returned by the buyers, did not contain the procured object, rather being replaced by a different device. There is a need to unobtrusively detect any such changes in the content of the sealed package. In this paper, the penetrability of Ultra-Wideband (UWB) radar is utilized to build a change detection technique for making a quantitative assessment of the presence of an 8″ Tab inside its sealed box, as opposed to other devices inside the same box. For this purpose, an electronic device is placed on an oscillating platform. A sliding Least-Of Constant False Alarm Rate (LO-CFAR) algorithm is applied on the radar scan data and the concept of detection sensitivity is introduced. Multiple experiments at different measurement instances are conducted to evaluate 1-class classification using detection sensitivity levels. These experiments show that the detection sensitivity levels provide a statistical basis for differentiating between objects. Also measurements were done at different distances, thus giving the result that the detection sensitivity level alone can be a determinant of the object type.