As one of the most successful optical correlation recognizers, hybrid optoelectronic joint transform correlator (HOJTC)
has received more and more attraction than the purely electronic way in the field of target detection and recognition. It
primarily because that HOJTC has the advantages of optics as well as those of electronics. This kind of combination
determines that the performance of HOJTC is closely related to optical configuration of system and digital image
processing technology. For the stability of optical part, a lot of efforts concerning image processing methods have been
made in recent years for improving the power of recognition of HOJTC. Edge contours play a decisive role in target
detection. In order to obtain adequate contour feature of target, the solution of edge extraction based on wavelet
multi-scale product is proposed. Normalized maximum and argument of each point could be defined utilizing wavelet
coefficient of image. Both of them contain the relation of coefficient product between each scale. Edge points
synthesized the information of multi-scale are extracted by searching local maxima along the direction of gradient. The
way adopted fully exploited the character of multi-resolution of wavelet. Simulation experiments and optical experiments
indicate that the energy of correlation peaks is obviously enhanced after the original image is processed by wavelet
multi-scale product, and it successfully realizes detection and recognition of infrared target.
Joint transform correlator is one of the most successful optical pattern recognizers which is composed of Ar<sup>+</sup> laser, spatial
filter, Fourier transform lenses, EALCD (electrically addressed liquid crystal displays) and CCD (Charge Coupled
Device). It has many applications in industry and military. The bottle technologies are how to recognize the target in
clutter scene and how to increase the brightness of the correlation peak which represents the detected target.
In this paper, we researched joint transform correlator (JTC) for infrared target detection, and used wavelet transform
technology to increase the contrast of input image which contains target and reference objects, reduced the high
frequency noise and realized target detection in clutter scene. According to our requirement, we selected the two-order
spline wavelet, and made input image have discrete binary wavelet transformed. Because edge always contains much
information of image, we adopted the method of modular maximum to extract the edge of different scales after wavelet
transform mentioned above. In addition, combined with threshold division which could eliminate some of the clutter, we
could get the remarkable edge.
As experiment result, we detected the infrared image, both with computer correlation matched method and optical JTC
method. The experiments show that the wavelet transform technology is one of the best data processing methods for the
target detection in clutter background.