Improving the performance of 1D object classification by using the Electoral College