Unlike a still camera, one cannot improve the resolution of an image captured by a video camera by increasing the length of the exposure; alternatively, increasing the density of pixels on the detector increases the noise.
An approach to improving the resolution of video camera images can be based on the use of algorithms that increase resolution by using spatial displacement. Even here, a substantial problem arises, namely, the introduction of motion blur that occurs when the camera is moved. This paper describes an interesting and successful attempt to circumvent these problems. The jitter camera constructed by the authors creates spatial displacement by moving the detector within the camera without moving the camera itself. The motion of the detector occurs while the detector is offloading the previous image, so that the detector is at rest again when the next image is to be taken. Besides this practical result, the paper analyzes the effect of motion blur on super-resolution algorithms. A lower bound on the deterioration due to motion blur is derived, showing that the algorithms cannot eliminate it entirely.
The authors also consider the performance of adaptive super-resolution for dynamic scenes and color images. The paper shows images that illustrate the effectiveness of the techniques and the camera. The authors indicate that they believe that their method is practical, and could be used in commercial video cameras. The paper is well written, and should be read by anyone with an interest in video imaging.