In view of the fact that the existing spatial indexes cannot meet diversified retrieval requirements, this paper proposes two new types of spatial indexes, which can search the spatial object with spatial attributes, text attributes and dividing attributes simultaneously. We apply classification techniques to the existing spatial object retrieval, and propose two spatial indexes including a hybrid index based on both dividing index and IR-Tree and first dividing index then IR-Tree index. Both spatial indexes not only meet the requirement of diversified spatial retrieval, but also solve the problem of the traditional spatial index's high maintenance cost effectively. The experiments based on the real data set show that the two indexes are effective and efficient. Compared with the traditional spatial indexes, the spatial indexes we proposed solve the problem of spatial object retrieval with dividing attribute, and achieve higher retrieval efficiency.