In this work, we propose a new approach for the connected component labeling task. The developed algorithm is based on a forward and backward scan of the image; from left to right and right to left and row by row. This approach takes into account the eight neighbors of the considered pixel. Unlike other algorithms that need to resolve equivalences in a second scan, with our approach we obtain at the end of the first scan an arranged table of resolved equivalent labels for each object. Another advantage of this method that a small number of labels may be required even for complicated objects. The second scan is done only to replace equivalent labels with the same representative label. These advantages make the approach more robust and sophisticated for many form recognition applications specifically for traffic signs identification which is our research subject. The software implementation of the method was firstly done using MATLAB and performances were evaluated. The simple formulation of the proposed method has enhanced the design of the VHDL circuit which is developed to be implanted on an FPGA card as part of a complete recognition system, and presented in this paper.