Genetic Algorithms: a stochastic approach for improving the current cadastre accuracies

The necessity for an analytical cadastre is impelled by the reality of the modern world. Over the past few decades or so, the issue of land management, including cadastral databases and information systems, has become increasingly acute. Even though, many countries still continue to rely upon a graphical, not homogeneous and inaccurate cadastre. This situation is far from ideal and is unsuitable for an efficient handling of land properties and real estate management. Much research has been done to improve the existing system; however, most currently employed techniques to achieve a digital cadastre, which are based on integrating old and new measurements, are mainly analytical, straightforward and aimed at resolving a specific situation rather than finding a comprehensive solution for reinstating the cadastral boundaries. A new unconventional approach that employs biological optimisation to attain unique, uniform and accurate coordinates under customary cadastral requirements - Genetic Algorithms (GAs) - is presented. This is a stochastic approach, originating in evolutionary algorithms, which is widely and successfully used in many other fields and disciplines. By mimicking biological processes, GAs offer an optimum solution obtained from a diverse range of possible initial solutions to a problem, by evaluating and evolving throughout a number of generations (iterations). The implementation of GAs principles in cadastral domain yielded good and promising results in a series of simulations performed on synthetic and real data. Based on these examinations it can be conclusively inferred that the GAs solution is more accurate than the conventional method - the coordinates are closer to their 'true' value than those obtained from the common Least Squares technique.