|Ph.D Student||Shnaidman Anna|
|Subject||Improvement of the Graphical Cadastre Based on Genetic|
|Department||Department of Civil and Environmental Engineering||Supervisors||PROFESSOR EMERITUS Yerach Doytsher|
|PROF. Uri Shoshani|
Cadastral measurement is a continuous process of recording the redefinitions and updates of boundaries that occur due to urban development and modern construction. A cadastral system is a vital factor in proper management of land properties and the overall economy of a country. Although the need for a computerized system is undisputed, an inhomogeneous cadastre of an analog nature is currently used by many countries, including Israel.
In order to employ technological innovations and respond to governments' needs, the existing graphical system requires re-engineering. A homogenous, accurate and accessible system needs to be established instead - the Analytical Cadastre system.
As an urgent problem, the transition to an Analytical Cadastre has given rise to much research. The majority of recently-developed academic works and government projects share a common ground: the principal practice is the Least Square (LS) method, which is an analytical approach directed at resolving a specific situation rather than finding a comprehensive solution for reinstating cadastral boundaries.
This research presents a new unconventional approach that employs Biological Optimization to attain uniform and accurate coordinates under customary cadastral requirements - Genetic Algorithms (GAs). This stochastic approach is used widely and successfully in many fields and disciplines. By mimicking biological processes and evaluating and evolving through a number of generations, GAs offer an optimum solution obtained from a diverse range of possible initial solutions to a problem.
In this research within the cadastral domain, the GAs were implemented as follows: each individual in the population represents a set of a mutation plan coordinates stored in an array (vector) structure; each vector contains all turning points within a given block with some coordinates mutual to several parcels; the data contains parcels with different areas, straight lines and pairs of parallel and perpendicular lines of unequal length and number of segments
At the first stage of this research, the proposed method was tested on a variety of synthetic data in order to evaluate its performance in terms of accuracy and validate its quality. Once the algorithms were carefully examined, they were implemented utilizing real data - parcellation plans. The specimens’ parcels form the cadastral constraints, whereas the geometrical ones are defined by straight and parallel lines (roads margins).
This innovative method provided very promising and more accurate results than the conventional LS technique, as may be conclusively inferred from numerous simulations and case studies analyses. Repeated simulation provided coordinate values very close to the initial, ideal coordinates and better than those obtained from LS solution. Moreover, results analysis showed standard deviation values of both ∆X and ∆Y coordinates of GAs solution 5 and 4 times smaller respectively, than those of the LS values. Examination of different Geometric and Cadastral constraints combinations show the GAs' ability to yield good solutions regardless of a given problem's features. Comparison of solution parameters between the conventional and stochastic solutions demonstrated a clear supremacy of the GAs method. Furthermore, the algorithm’s enormous potential in providing a good solution in the analytical cadastre sphere was made evident.