Radio maps, as an effective tool for data visualization, play a crucial role in displaying the performance of multivariable data across various dimensions. Especially in modern data analysis, radar maps offer an intuitive means to compare and analyze complex datasets. However, physical, weather, and terrain constraints often result in the inability to obtain measurement values in certain areas. The non-uniform characteristics and limited observations further complicate the reconstruction of radio maps. This paper introduces an innovative exemplar-based approach for restoring and filling radio maps in the absence of observational values. A novel propagation priority calculation that utilizes Power Spectral Density (PSD) patterns and radio characteristics grounds our method to guide the predictive filling process in a logical sequence. We also propose two schemes to reconstruct the maps effectively using their observable parts. The test results demonstrate the effectiveness of our proposed methods in radio map reconstruction. Additionally, to address the significant limitation of the current scarcity of open radar map data, we developed the BRAT-Lab Radiomap open-source dataset generated through high-fidelity simulation techniques. This dataset encompasses radio maps of multiple frequency bands and varying resolutions. This dataset facilitates academic research and provides substantial support for practical wireless network design and optimization.