Guarding a surface is a well known optimization problem of the visibility site analysis and has many applications. The basic problem is searching for the minimum number of guards needed to guard (see) the entire surface. More realistic is the guarding where the number of guards is upward limited and the optimization problem is to search for their locations in order to guard as much sur- face as possible. In the paper this problem is treated in de- tails. Several known heuristics (greedy add, greedy add with swap and stingy drop) are re- vised and a new technique called solution im- proving technique is proposed. The technique improves the results of the known algorithms and is used in indirect solving of the problem. Tests on 44 DEMs from USGS DEM Reposi- tory showed that our technique yields compar- ative results for smaller number of guards and better results for higher number of guards.