Sensor database systems serve the applications by querying the sensor nodes in a sensor network in ways facilitated by a query language. Spatial queries satisfy the needs of many applications which need data from different spaces in the sensor network. However, the existing sensor databases have limitation of expressing irregular-shaped space in a query and processing it efficiently. Therefore, we propose target space assignment facility and processing scheme as an improvement of our previous sensor database query language SNQL (Sensor Network Query Language). It enables users to specify more accurate target spaces of sensor nodes by intersect, union and minus operations in addition to specifying a simple space expression. Also, our proposed scheme supports selective query dissemination to polygon-shaped spaces. We evaluate the efficiency of our proposed target space assignment query processing scheme with other sensor databases. We measured the number of transmission count, node count for the query and non-target node rate by changing the number of vertices of a space and the number of spaces. The result shows our scheme provides an efficient way to process spatial query in sensor database.