This paper investigates an environmental search problem using a multi-robot system to search small objects. Specifically, the system divides the SLAM-generated map into rooms and then allocates these rooms to individual ground robots to ensure that the full-coverage search process is completed in the shortest time without interference between robots. The primary innovation of this paper is the proposal of a room region processing algorithm based on the Doubly Connected Edge List (DCEL) data structure and a room allocation algorithm based on Mixed-Integer Linear Programming (MILP), which together form the processing and allocation method. Numerical experiments demonstrate the effectiveness of the processing algorithm, the balance, safety, and optimality of task allocation, as well as its applicability to multi-speed ground robot systems. Additionally, the effectiveness of both the processing and allocation algorithms was validated through physical experiments.