Here is an excellent business opportunity. The US Defense Department is looking for a smart camera that takes clear pictures even when you shake it. If you have ideas on how to make it, they will pay $850,000 for it. Any takers?

The DoD uses digital cameras on unmanned vehicles and at remote stations for long-standoff surveillance and reconnaissance applications. The images acquired by these cameras are subject to apparent motion of the field of view due to vibration of the camera as well as motion and distortion due to atmospheric effects. For many applications, stabilization of the image is highly desirable prior to transmission from the camera to an observer.

Image stabilization can be achieved to varying degrees by a multitude of methodologies, some of which have found success within the commercial still and video camera industry. Mechanical image stabilization actuates the imager in order to suppress jitter present on the observers platform from inducing blurring or perceived motion within the image. Digital image stabilization techniques shift the image data spatially within the output array to compensate for apparent motion of the field of view. These digital methods typically employ scene based approaches to deduce own motion and have demonstrated sub-pixel levels of jitter-rejection given a scene with stationary segments that can be tracked and registered. These scene conditions cannot always be guaranteed under poor visibility or low-light scenarios or when the scene is dominated by moving objects. As a result, scene-based stabilization approaches are explicitly excluded in this topic. Instead, the preferred solution equips the camera itself with instrumentation to sense its own motion and stabilize (either digitally or mechanically) the image before transmission to the observer.

Both Feeback Inertial Image Stabilzation and Feedback Inertial Image Stabilization are established approaches to inertial camera stabilization. However, this DoD wants innovative solutions that push beyond existing performance limits and achieve new levels of stabilization performance required by long-standoff observation systems. Long standoff distances (multiple kilometer paths) and narrow fields of view will entail sub milliradian (10s of microradian) levels of stabilization. This is in the presence of 10 – 100 milliradian levels of local vibrational motion of the camera over frequencies from 1 to 100 Hz. The levels of motion and desired stabilization motivate new approaches to inertial sensing and corresponding advances in embedded processing to digitally correct for measured motion.

At the same time, new levels of smart camera integration are required to seamlessly incorporate the desired image stabilization into fast-framing low-light applications. The camera must be smart to enable easy integration into existing video surveillance applications. All processing must be local within the camera device, without imposing external computational or infrastructure requirements. Typical applications could involve mounting the camera on an unmanned ground vehicle or on a surveillance tower. The camera must be capable of providing better than VGA levels of resolution at greater than 60 frames per second. The ability to provide color images in low-light (sub milli-LUX) is critical to a variety of surveillance missions for discrimination of objects under difficult observational conditions. It would be highly desirable to integrate the image stabilization into an embedded processing capability that is scalable and reconfigurable in order to accommodate additional image processing algorithms for object discrimination and image compression. However, the development and implementation of these algorithms is not the intended focus of this topic.

Here are some references to get you started:

1. R. A. Gross, “Analysis of Internal-Inertial Image Stabilization,” Appl. Opt. 10, 1422-1431, 1971.

2. Peter Corke, Jorge Lobo, and Jorge Dias, An introduction to inertial and visual sensing, The International Journal of Robotics Research (IJRR) Special Issue from the 2nd Workshop on Integration of Vision and Inertial Sensors., 26(6):519 V535, June 2007.

3. Morimoto, C. and Chellappa, R., Evaluation of image stabilization algorithms, Acoustics, Speech and Signal Processing, 1998. Proceedings of the 1998 IEEE International Conference on Volume 5, 12-15 May 1998.

4. Tico, M. and Vehvilainen, M,, Robust method of digital image stabilization, Communications, Control and Signal Processing, 2008. ISCCSP 2008. 3rd International Symposium, 12-14 March 2008

5. Scott W. Teare; Sergio R. Restaino, Introduction to Image Stabilization, SPIE Tutorial Texts in Optical Engineering Vol. TT73

6. PTU-DISM – Inertial Stabilization Module – www.dperception.com/products_family_advanced-features-ism.html