Genetic programming: an introduction: on the automatic evolution of computer programs and its applications
Latin hypercube sampling of Gaussian random fields
Technometrics
| Hi-index | 0.00 |
Three new methods are developed to generate neutral spatial models for pattern recognition on raster data. The first method employs Genetic Programming (GP), the second Sequential Gaussian Simulation (SGS), and the third Conditional Pixel Swapping (CPS) in order to produce sets of "neutral images" that provide a probabilistic assessment of how unlikely an observed spatial pattern on a target image is under the null hypothesis. The sets of neutral images generated by the three methods are found to preserve different aspects of spatial autocorrelation on the target image. This preliminary research demonstrates the feasibility of using neutral image generation in spatial pattern recognition.