The automation level of bar counting is one of the important aspects for measuring the intelligence and informatization degree of enterprises. A fast bar counting method based on object features extraction was proposed to solve the problem of rapid counting of stacked bars in production and life according to the distribution characteristics of target pixels in the binary image. The efficient connectedcomponent labeling algorithm was use to label the object pixels in the bar binary image and count the object feature value. According to the object feature value, the automatic calculation of stacked bar was carried out. The bar image only needs to be scanned one time during the entire processing. The experimental results show that by using the proposed method, the automatic counting of stacked bars can be realize quickly and accurately.