Minimum norm solution based approach to decoding of real number BCH codes

In this paper we consider the special case of the underdetermined LS problem when the difference between the number of unknowns and the number of equations equals one. We propose a new method to improve the minimum norm solution based on using the estimate of the norm of the exact solution. The method is applied to the problem of syndrome based error control image coding over real fields. A series of computer simulations show the significant gain in output signal to noise ratio at high bit error rate in the channel