Can text objects in the layout be used to improve device recognition in LVS?

Version 1

    Strictly speaking, the LVS device extraction process does not look at text, the process is guided solely by the interaction of shapes on the pin layers with a shape on the seed layer.  However, derivation of the seed layer shape may involve text.  For example, consider the following two shapes and text objects:

    ----------------------------              --------------------------
    |                          |              |                        |
    |           ?MN?           |              |          ?MNX?         |
    |                          |              |                        |
    ----------------------------              --------------------------

    Both of the above shapes are on layer pwell and each interacts with a text object (the text objects are ?MN? and ?MNX?).  The following SVRF statements use the text objects to derive needed seed layer shapes for the two device types (MN and MNX):

     

    mn_seed = pwell WITH TEXT ?MN?
    mnx_seed = pwell WITH TEXT ?MNX?

     

    The two derived layers, mn_seed and mnx_seed, can now be used in two DEVICE statements to specify these two device types.