In recent years we are faced with the growing trend of illegal buildings (IBs) construction in urban areas as the urban population and expansion have been significantly increased. Unfortunately, some of the building constructors commit some building infractions which may have irreparable consequences for urban structures. Lack of awareness of some of the constructors from the building construction laws and regulations, the high costs of getting building license from the municipality, the willingness of some of the constructors to construct more than allowable building density, are some of the main reasons of building infractions. Providing some new and appropriate solutions of detecting and monitoring building infractions is essential to prevent the continuation of IBs construction. Field check, as the currently used method of IB detection in Iran is time, cost and man power consuming and it may be ended in some collusion between the municipality inspectors and constructors. In addition, using this unplanned and irregular method of illegal building monitoring may be ended in untimely detection of illegal buildings. Acquiring multi-temporal satellite images and using image processing techniques for semi-automatic change detection is one of the optimum methods which can be used to detect the suspected IBs.

In this research, a semi-automatic method of IB detection based on the pixel wise fuzzy XOR operator change detection, using bi-temporal satellite images of west of Tehran, the urban maps at the scale of 1:2000 and an up to date municipality property database containing building information, is proposed. In the pre-processing step, firstly the radiometric and geometric corrections have been undertaken to the images. Then, the changed pixels are detected using a fuzzy XOR classification method. High accuracy and speed of the mentioned fuzzy method are the main reasons of taking advantage of this method. The change percentage of each building is obtained by calculating the ratio of the changed pixels to the total pixels of each building. The change percentage of each building is compared with the standard change threshold of the study area (allowable building density (60%)) and consequently the buildings which are under construction are detected. In the next step, the IBs are detected as a suspected point undertaking a query to the municipal database. Finally, the existence of authenticity of IBs is checked undertaking a field visit from the suspected point.

The results verified that from the 343 buildings appeared in the image, 21 buildings were detected as under construction and 4 buildings detected as illegal buildings, respectively. Furthermore, the overall accuracies of 92%, 83% and 100% were obtained for the fuzzy-based classification, detection of under construction buildings and IBs detection, respectively.