To process the massive quantities of information received from available infrared focal plane arrays, Celnet Technology Corporation proposes to design an on-focal plane processor architecture suitable for infrared applications. The proposed system utilizes Cellular Neural Network Universal Machine (CNNUM) chips for processing, and analog RAM for fast cache. The advantages of using CNNUM technology include improved data relevancy at higher frame rates, higher pixel counts, low power, and compact design plus the possibility of near- sensor implementation. Phase I work will include the design of the CNNUM VLSI chip, the analog RAM chip, and system integration. A number of algorithms will be simulated to demonstrate and validate the proposed system for infrared image processing.