The real-time image recognition based on joint transform correlation

The technology of image recognition is used widely in many application fields. An photoelectric hybrid and real-time image recognition experimental system based on the joint transform correlation is designed and built in the paper. The electro-addressed spatial light modulator transforms digital images to coherent light images and the optical device processes quickly the two-dimensional image correlation operations by parallel processing in the system, then the power spectrum of joint transform is detected by the square law detector CCD. The electrical device of the recognition system cooperates reasonably with optical device due division of functions and completes the real-time image correlation recognition. In order to improve the correlation signal characteristics, the technology of power spectrum amplification and high-pass filter are used to process the power spectrum, which not only reduces the spatial correlation dimension of CCD detector, but also weakens the central zero-order signal, sharpens and enhances the first-order cross-correlation signal characteristics and improves the signal-noise ratio (SNR) of correlation signals. The recognition experiments of real-scenery images are done based on the system and the experimental results are given.