A sequential cycle cascade correlation system for wavelet transform (WT) of images and correlation recognition based on feature extracting by WT is presented. The basic structure of the correlation system consists of a usual optical correlator. In this correlator, spatial light modulators can be used as real time input device. A piece of photorefractive crystal is used as an information storage element, a vast number of signal data are recorded in Fourier transform holograms by spatially encoding reference beams, such as different spatial frequency spectra of object functions, wavelet functions and WT functions of signals. The input and output signals, and sequential operation procedures of the correlation system are controlled by a computer, which can collect the input and the output signals by CCD cameras, record them, and process them in nonlinear thresholding. In this paper, the optical analogous experimental results of wavelet transform of 2-D images and their correlation recognition are provided in the cases of 1-D and 2-D Haar wavelets. For comparison, the optical result of usual correlation experiment is also presented. It is very obvious that the wavelet transform correlation recognition possesses more high discriminability than the usual correlation recognition. In addition, some other advantages of the cycle-cascade correlation system are discussed finally.