Real-time Object Recognition and Classification by Code Division Multiplexed Phase-only Encoded Filters

Abstract

A multiobject shift invariant pattern recognition system that uses Code division multiplexed binary phase-only filters is presented. The system computes the binary correlation between an input pattern with a generalized set of pattern functions. This technique uses a filter which consists of a set of binary phase-only code division multiplexed reference pattern functions. The functions in the reference set may correspond to either different objects or different variations of the object under study. In the proposed architecture, the position of the output binary correlation peaks represent the locations of the objects in the input plane. This is different from phase coded pattern recognition systems in which the location of the output peak determines the class of the input object+. Here, the system is shift invariant and multiple objects can be present at the input plane. The pattern recognition is performed without increasing the space-bandwidth product requirements of the system. Computer simulations of the architecture are also used to study the performance of the pattern recognition architecture. The signal-to-noise ratio (SNR), and the peak to the maximum correlation sidelobe ratio (P/SL) are evaluated as the criteria for the system performance.