While the value of image processing has been longly recognized, this is usually done during post-processing. For scientific application, the presence of large noise errors, data drop out, and dead sensors would invalidate any conclusion made from the data until noise-removal and sensor calibration has been accomplished. With the growing need for ruggedized, real-time image acquisition systems, including applications to automotive and aerospace, post processing may not be an option. With post processing, the operator does not have the opportunity to view the cleaned-up image. Focal plane arrays are plagued by bad sensors, high manufacturing costs, and low yields, often forcing a six digit cost tag. Perhaps infrared camera design is too serious an issue to leave to the camera manufacturers. Alternative camera designs using a single spinning mirror can yield perfect infrared images at rates up to 12000 frames per second using a fraction of the hardware in the current focal-plane arrays. Using a 768 X 5 sensor array, redundant 2048 X 768 images are produced by each row of the sensor array. Sensor arrays with flawed sensors would no longer need to be discarded because data from dead sensors can be discarded, thus increasing manufacturing yields and reducing manufacturing costs. Furthermore, very rapid image processing chips are available, allowing for real-time morphological image processing (including real-time sensor calibration), thus significantly increasing thermal precision, making thermal imaging amenable for an increased variety of applications.