Model decomposition via optical delay estimation using the camera tracking in an active noise control system for creation of a movable quiet zone

This article presents an adaptive control system for active local reduction of broadband noise at moving locations, using the camera tracking for path model decomposition to reduce the filter lengths. To enable a dynamic placement of local quiet zones with respect to a moving target, previous works proposed the combina- tion of the remote microphone technique with a camera-based target tracking. The estimated 3-D coordinates are used at run time for a dynamic update of the trans- fer paths models, which correspond to virtual microphones at predetermined loca- tions. Nevertheless, whereas current applications like headrests feature only short distances to the target, the considered ANC system aims at larger ranges for the moving quiet zone, resulting in long impulse responses and high delays in the transfer paths. To reduce the associated model lengths, this work evaluates a decomposed identification stage that takes advantage of the camera-tracking sys- tem by estimating the acoustic path delays optically. In experiments, this method is shown to be superior to signal processing-based approaches like cross-correlation and thus enables an accurate separation of the delays. Eventually, the valida- tion of the broadband ANC performance at a moving microphone shows that the quiet zone could be tracked over a range of 0.8 m, while using only few transfer path models and saving over 100 coefficients due to the proposed decomposition.