Abstract: In this paper, taking the maximum absolute error for vector data into account, we propose an approximate DCT compression method, which is designed for controlling the maximum absolute error and the high arithmetic complexity, aimed at traditional local compression algorithm for vector data based on discrete cosine transform. First of all, we construct vector data blocks on the basis of vector data topology. Secondly, we calculate the optimization solution of the agreed matrix in accor-dance with the orthogonality of approximate DCT transform, and then set the solution which has the minimum computational complexity as the conversion matrix for approximate DCT transform to ma-ximally guarantee the precision of vector data. Finally, we combine the total energy between the DCT transform and accurate approximate DCT transform of vector data with maximum absolute error of the reconstructed data, and use three spline interpolation function for data which exceeds thresholds of error to ensure maximum accuracy. Experimental results show that the proposed method has low computational complexity and high speed compression, it can keep the topological relation of the spatial data and the accuracy of the data while reducing the compression rate.