中文摘要：

为了提高激波和旋涡等复杂流动现象的捕捉精度和计算效率，基于图像处理的思想，研究了非结构网格的自适应加密和稀疏算法。将流场计算网格与图像像素类比，借鉴图像处理的方法，提出了一种流场自适应区域的分割算法。该方法能自动对自适应区进行分层，中间区域网格可以达到原来多次自适应的效果，而且提高了生成网格的质量。将算法嵌入到流场求解器中，根据流场求解结果对网格进行自适应，并使用算例进行了验证。

英文摘要：

Solution-based grid adaptation is a subject of great significance to the practical application of unstructured grid in CFD simulations, especially in the simulations of complex flow phenomena. The paper presents an efficient solution-adaptive procedure, based on the image process methodology. It regards the grids as the analogical pixels of picture, and the values of flow parameters （density, velocity, etc） as the gray. The image process techniques are employed to isolate the regions in the field which contains physical features of interest. These regions are reduced to several groups of simple geometric entities representing the physical features via a feature detector function, which contains different values. Therefore, the transition region of adaptive grid between the finer and initial regions will be generated automatically after the adaptive procedure, and the quality of the adapted mesh is ensured. The refinement and coarsening techniques, implemented based on the values of the feature detector function, are developed as well. The procedure was coupled with a flow solver. Results were presented for a variety of CFD applications. The method is successfully shown to be effective for aerodynamic applications, detecting regions of interest, etc.