Particle swarm optimization (PSO) is a new promising evolutionary algorithm for the optimization and search problem. One problem of PSO is its tendency to trap into local optima due to its mechanism in information sharing. This paper proposes a novel hybrid PSO, namely (HPSO) technique by merging both a mutation operator and natural selection to solve the problem of premature convergence. By introducing Cauchy mutation and evolutionary selection strategy based on roulette wheel selection, HPSO could greatly reduce the probability of trapping into local optimum. HPSO is proposed to improve the performance of fragile watermarking based DCT which results in enhancing both the quality of the watermarked image and the extracted watermark. After embedding watermark to the original image in the frequency domain, the conversion of real numbers of the modified coefficients in frequency domain to integer numbers in spatial domain produces some rounding errors problem. This problem results in completely different of the extracted watermark from the embedded watermark. The new developed PSO with evolutionary operators is carried out for correcting the rounding errors by training a translation map used to modify the inverse DCT (IDCT) coefficients from real to integer numbers. The experimental results show the superiority of the proposed algorithm comparing with the standard PSO for improving the performance of DCT fragile watermarking. Besides, it has been shown that the developed PSO is faster in convergence and the obtained results proved to have higher fitness than the other algorithm.
Particle Swarm Optimization, Evolutionary Operators, DCT Fragile Watermarking.