This paper introduces a novel road extraction algorithm in two stages of road detection and road vectorization. In the road detection stage, road class image is obtained using fuzzy C-means clustering and some post processing operations. In the vectorization stage road key points on the road centerline is obtained by an innovative approach of dynamic road pixels clustering using particle swarm optimization. The proposed algorithm is able to automatically optimize number and position of road key points without considering the prior information about the initial number and position of cluster centers by designing a new cost function. The optimized road key points were connected using weighted graph theory. Different high resolution images of Ikonos in urban, non-urban, and mountainous areas were tested and several quality measures including RMSE, correctness, completeness, and quality were calculated. Extracting different road shapes with RMSE less than 1.3 and quality greater than 0.86 in different areas proves the efficiency of the algorithm in yielding complete road networks.

In this paper an efficient method for automatic road detection from high-resolution multi-spectral IKONOS images is presented. The system includes four main steps: In the first step the input image is segmented into road and background classes using K-means clustering and then some misclassification pixels in road binary map are removed using a median filter. In the second step, angular texture shape descriptors (mean, compactness and eccentricity) are driven for every road pixel in road binary map. In the third step, these descriptors are introduced into a fuzzy inference system. In the fuzzy system each descriptor is introduced as a linguistic variable with Gaussian membership functions while their parameters are set automatically according to statistical properties of each descriptor. Also, some fuzzy if-then rules are established. By using the centroid defuzzification, road network is distinguished from other spectrally similar classes (shadows, buildings, parking lots and etc). Then, road network is completed by connecting road pixels together and removed of small paths. In the last step of system evaluation, obtained results are compared with manually extracted road network and some accuracy assessment parameters are computed. The conventional maximum likelihood classification (MLC) is also implemented and the same accuracy assessment parameters are determined for comparison. Preliminary results show the effectiveness of the methodology of this paper in both resembling the desired results of road networks and achieving a good automation level. Furthermore, it outperforms MLC to high extent. .

Orbital parameters model is one of the fully constrained physical models for geometrical correction of satellite imagery. The model has been developed to cover the physical conditions prevailing in the acquisition period of satellite platforms. The multiplicity of parameters in the modeling and correlation between them causes difficulties in solving the system of equations. Generally, problems are greatly reduced by adding the model parameters in the form of quasi observations and controlling trend of corrections by the values of each parameter's weights. However, it is difficult to determine the correct values of quasi observations' weight due to the approximate precision of auxiliary data. Furthermore wrong weights of quasi observations impose additional parameters in the model structure. In this article, by providing some scenarios, the level of correlations between orbital model parameters and also the capability of rigid orbital model in the covering effect of perturbations are evaluated. The obtained results proved that unconsidered perturbations, occurring in a specified domain, are coverable by other correlated parameters of rigid orbital parameters model. This ability of the rigid orbital parameters model is more evident when quasi observations are not applied in the model.