A new scale-invariant correlation filter based on synthetic-discriminant-function (SDF) was designed. In this filter, the correlation output function modulated by SDF in frequency-domian was quantized to complexes of a quaduploe to obtain sharp correlation peak. To satisfy the equal correlation peak criteria , an iterative algorithm was utilized to modify the coefficients of SDF. Thus scale-invariant recognition was achieved. Computer simulation results show that the proposed method performs better compared with the basic SDF and the binary phase-only SDF(BPOF SDF) for the target images from training set.