The task of identifying corresponding objects between different geospatial datasets is known as matching problem which has variety of applications, e.g., conflation, spatial data enrichment, updating, change detection, and quality assessment. Matching problems in vector data models can be divided into three categories of point, linear, and polygon problems according to the type of geometry are being used in matching process. Furthermore, the similarity measures utilized in order to calculate the degree of similarity between two objects can be classified into three groups of semantic, geometric, and spatial relation measures. Since there are few studies of polygon matching problems compared to linear objects and geometric measures are easily accessible, among these various kinds of matching problems, the purpose of this study is to propose an approach to identify corresponding polygon objects based on geometric properties. The proposed approach contains four stages of preprocessing, spatial similarity calculation, extraction of corresponding relations, and results analysis. The preprocessing stage consists of creating uniformity among data formats and coordinated systems, and also removing topological errors. In the similarity calculation stage, a probability based matching algorithm is presented in which the four similarity measures of distance, overlapped area, orientation, and shape are being used. Then, the six kinds of corresponding relations including 1:0, 0:1, 1:1, 1:N, N:1, and N:M relations are obtained as the result of similarity calculation stage. At the end, the results are analyzed through evaluation of the algorithm. Besides that, the impacts of each similarity measure, solitary and in combination with other measures, have been studied in the final precision of algorithm as the evaluation process. The implementation is carried out on the district 6 of Tehran city as the case study area by using two different datasets at the scale of 1:2000 and 1:25000. The evaluation of proposed method has been achieved according to three criterions of Precision, Recall, and F1-score. Also the manual matching of two datasets is needed to evaluate the proposed algorithm. The results show that the proposed algorithm by using all four similarity measures has reached the F1-score precision criteria of 99% which is quite high over the case study area. Furthermore, the influence of each similarity measure has been studies both solitary and in combination with other measures which shows that the precision is not necessarily increased by the increase in the number of similarity measures. As an illustration, the exclusive usage of overlapped area measure has far higher precision in compared with the utilization of two measures of distance and orientation. Consequently, in order to decrease the cost and time of processing, it is better to use the least number of similarity measures which positively affect the algorithm precision. Also the precision of proposed method was compared to one of the latest work of polygon matching problems using geometric measures.  The results demonstrate that the precision of polygon matching problem has been improved compared to the previous work.