There are two drawbacks for the traditional LIC (line integral convolution) methods to visualize data of unstructured vector field: one is its restriction to a fixed resolution and the other is its poor ability of revealing vector magnitude. In this paper, a new method is presented for computing LIC images with unstructured vector data. In this method, numerical integrators are utilized to allow computation at arbitrary resolution, and a mask method is employed for masking vector data by the region of DDA curve. This method can reduce computational cost by minimizing the total number of DDA lines. Because the size and shape of a DDA region are decided by the vectors' magnitude on the DDA line, as a result, in LIC images the vectors with big magnitude are surrounded with dense DDA lines while the vectors with small size are surrounded with sparse DDA lines. The inner structure of the field as well as the magnitude of vectors can be depicted more clearly compared with the traditional LIC.