This thesis presents a simple binary filter method for feature extraction and fault enhancement of fault attribute data. Fault attributes are commonly used in structural interpretive studies to detect faults. However, they also tend to detect stratigraphic discontinuities and noise, and this provides a need to remove unwanted features and to sort out important information. This has been the motivation behind this thesis. Structural geology, seismic interpretation and data processing have been combined to develop the presented methodical approach. This approach involves converting the fault attribute data to binary data, as well as assuming that each individual binary object has a set of properties that represent fault properties and can be filtered. Each binary operation has been through an iterative process of testing and evaluating different parameters for the most optimal use, and the procedure has further evolved through this testing. All computational operations have been executed in MATLAB r2014A and results have been evaluated subjectively in Petrel 2015. Finally, development and application of seven binary filters is presented. They all in some way measure properties of binary objects in two- and/or three-dimensions and they all, to some extent, enhance or extract structural elements or trends. The specific attribute data are fault likelihood coherence data derived from seismic data with a semblance-based coherence algorithm. All completed filters are developed specifically for this fault likelihood coherence data. However, it is assumed that other fault attribute data could be used after some adaption. The original seismic data is acquired in the southwest Barents Sea.