A technique of encoding an extended incoherent source for image subtraction is presented. The source encoding is obtained from the coherence requirement for image subtraction operation. Since the coherence requirement is a point-pair concept for image subtraction the encoding can take place by spatial sampling an extended incoherent source with narrow slit apertures. The basic advantage of the source encoding is to increase the available light power for the processing operation, so that the inherent difficulty of obtaining a very small incoherent source can be alleviated. Experimental results obtained with this encoded incoherent source are given. Comparisons with the results obtained by processing technique are also provided.