Smart Image Processing

As digital cameras are widely used, the way to store image is changed from analog type, e.g., film, to digital type, e.g., computer data. Because these digital images are expressed by certain numbers, various mathematical algorithms can be applied to them. Image processing is a process on this digitalized image for various purposes. For example, there is a selective elimination filter which can remove like snow, rain and bugs and bring clean video. Also, an enhancement filter can increase the saturation value in a hazy video and, video stabilization can compensate camera shaking when a camera is swayed by unintentional forces, such as wind. These techniques play an important role individually, and even be a pre-processes before the adaptation of upper level algorithm.

Figure 1. Result of the Video Stabilization

Figure 2. Result of the snow, rain and bug removal fliter


When these algorithms are used as pre-processing and combined with the upper level algorithms, they can increase the computational complexity. In addition, the quality of video or the performance of upper level algorithms can be degraded if filters or stabilization process is applied when they are unnecessary. For this reason, filters should be adopted when it is required actually. Smart image processing is a solution for this problem. By smart image processing scheme, the system itself analyze the video and determines automatically whether the filters or stabilization should be applied or not in certain situations. But, automation procedure should be done carefully and it is required to prevent malfunction. For this reason, the selection measurement for decision is important and there are many on-going researches in this area.

Figure 3. Result of the automatic filtering