Noise Removal and Registration of Remote Sensing Aircraft Data for Energy and Mineral Resource Studies

Susanne Hummer-Miller

Both noise removal and accurate registration of remote sensing aircraft data are important for energy and mineral resource studies. Accurate geometric registration to a map base is required for studying spatial associations of anomalies, for accurate field checking of results, and for comparison with other data sets. Noise may be present due to the instrument and/or the aircraft motion. Innovative technique for noise removal and registration were applied to thermal infrared multispectral scanner (TIMS) aircraft data of the Carlin gold deposit in northeastern Nevada and McClure Mountain in central Colorado. The techniques presented are applicable to any remotely sensed data set and are particularly useful for aircraft data.

Registration of aircraft data to a map base is usually unique for every acquisition site due to the motion of the platform. The registration technique starts with registering Thematic Mapper (TM) satellite data of the site to a topographic map base using an affine transformation. Subsequently, aircraft data, ideally acquired under illumination conditions similar to TM, are registered to the TM; mapping pairs are selected and a transformation that uses the weighted sum of the nearest mapping pairs is applied.

Noise in aircraft data can be of many types. Random and bit-error problems can be removed by applying a median threshold filter in the spatial domain. The remaining noise can be diminished by creating special filters in the Fourier domain using a trial-and-error process.

Both day and night data from the two example sites exhibit considerable noise and geometric distortion. The accuracy of the registration technique is illustrated by superimposing digital topographic data and digital geologic unit data onto images before and after registration. The effectiveness of the noise removal is evaluated before and after noise filtering using decorrelated images that enhance noise. The geometrically registered, noise-filtered data have been used to extract subtle differences in the emissivity contrast of rock-forming minerals at the two sites.

AAPG Search and Discovery Article #91022©1989 AAPG Annual Convention, April 23-26, 1989, San Antonio, Texas.