Data Adaptive Denoising of Magnetic Anomaly Maps

In the past one to two decades, a number of data adaptive denoising tools have been proposed in the image processing community. The basic idea of these filter approaches is to establish the filter weights by considering the actual sampling values, their local statistics and similarities. This helps to minimize image blurring and to preserve edges and corners. As such filter characteristics are also desirable for noise attenuation in near-surface magnetic data sets, we propose to adopt these methods for processing magnetic anomaly maps collected across archaeological targets. Here, we test and evaluate two selected methods (a generalized Kuwahara-style filter and the steering kernel method) to denoise a magnetic data set collected across Neolithic ring structure in Germany. Our results show that both methods are successful in removing prominent noise features present in our data. Concurrently, they largely preserve local structures; i.e., blurred images as typically observed after applying filters using a fixed filter mask are avoided. Thus, the methods can be considered as promising and novel approaches for denoising magnetic data sets.