Need for Image Processing in Infrared Camera Design

摘要

While the value of image processing has been longly recognized, this is usuallydone during post-processing. For scientific application the presence of large noise errors, data drop-out, and dead sensors would invalidate any conclusions made flom the data until noise-removal and sensor calibration has been accomplished. With the growing need for ruggedized, real-time image acquisitions systems, including applications to automotive and aerospace, post processing is not an option. In some instances, 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 768x5 sensor array, redundant 2048x768 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.