The feasibility of using low-resolution thermal imagers for home security applications is analyzed, taking low cost as the primary consideration. The smallest possible sensor size and resolution are cho- sen as the operative criteria and derived by simulation, in accordance with the optical constraints of a general home security system and the minimum target feature recognizable using image processing. Low- resolution simulated thermal images were generated by downgrading the high-resolution images captured by an uncooled IR camera, through sampling and modification. Caricatures of human beings and family pets are extracted for recognition using aspect-ratio and neural network meth- ods, which are compared with one another for detection probability. It is demonstrated that highly reliable detection of human beings or pets with a minimal target feature of 838 pixels can be obtained using the neural network method. Also, a fire can be detected early using its temporal size variation and higher temperature. Finally, low-cost fabrication of the proposed low-resolution passive infrared imaging system with an un- cooled FPA sensor utilizing a fully standard application-specific IC CMOS process is also discussed in detail. © 2000 Society of Photo-Optical Instrumentation Engineers. (S0091-3286(00)02706-9)