Evolutionary Computation for Remote Sensing Applications
Greg Easson, Henrique G. Momm · Geography Compass · 2010
Abstract As the volume of available remotely sensed imagery increases so does the need to extract more specific information in a timely and cost‐effective fashion to enhance and/or update decision support systems. This manuscript provides an overview of the existing image information extraction techniques using evolutionary computation algorithms. Emphasis is given to remote sensing applications. The literature investigated is further divided into four groups based on the research objectives: image enhancement, image classification, feature extraction, and modeling. Some of the strengths of evolutionary computation are discussed, such as robust classification capabilities and ability to work with linear and non‐linear problems. Conversely, the works reviewed reveal some of the challenges in the integration of evolutionary computation and remote sensing as well as possible solutions. Limited work was found addressing the multi‐class image classification problem and assessment of the randomness involved in the generation of the first set of candidate solutions from which the algorithm begins. Based on our study, the use of evolutionary computation algorithms for information extraction from imagery is likely to grow, specifically in two areas: inverse modeling and object‐based approaches. This technology is still in its early stages of development and the capabilities of evolutionary computation are well suited to the complexities of the image processing and information extraction problems.