In order to make full use of the spatial-spectral bandwidth of optical correlation recognition system and to enhance the parallel processing efficiency and recognition accuracy of optical correlation recognition technology, a recognition method for multiple channeled joint transform correlation is proposed based on the compression and translation of power spectra, in which the standard deviation of the variation of position of correlation peak is derived to act as a new criterion for correlation recognition. The scene image and N reference images are first uploaded on the different zones of the input spatial light modulator. Then, the optimized phase maps using the iterative algorithm are superimposed onto the images. In the Fourier plane, interference will occur between the Fourier spectrum of the scene image and that of every single reference image within different zones of the Fourier plane. No interference of Fourier spectra among the reference images will appear when the restriction parameter is suitably adjusted in the phase optimization algorithm. As a result, the N channeled parallel processing without crosstalk is achieved. The relationship between the localized peak clutter mean of the Fourier spectrum with the optimized phase and the standard deviation of the variation of position of correlation peak is analyzed and utilized as a criterion for the selective preference on the optimized phase mask. The results indicate that the proposed multiple channeled joint transform correlator can achieve 16-channel correlation recognition without the increase of system complexity compared with the classical optical correlator under specific experimental conditions, which is quite important for the practicality of an optical correlator.