Recovery techniques for images and videos with missing data in disaster situation

Summary form only given. This research aims at developing recovery techniques for images and videos with missing data due to instability of communications networks in disaster situations. In heavy disaster situations like the Great East Japan Earthquake, instability of communications networks causes packet loss or bit error. Then, some data of transmitted images (videos) are lost and thus there appear regions without information, which are called missing data. Missing data in an image will appear in rectangular regions shown in Fig. 1(b) because block coding is generally used. Recovery techniques for images (video) with missing data need not to provide high quality images, but to provide medium quality images without break. This demands low power recovery algorithms with fewer computational burdens. Recovery problem for images (vidoes) with missing data is regarded as one of noise reduction problems in the fields of signal, image and video processing. Many noise reduction techniques deal with relatively small amplitude additive noise to each pixel in an image, but missing data are considered to be large amplitude noise appeared in rectangular regions because of their nature of information loss. Thus, nonlinear signal signal processing (median filtering, morphological filtering [1], and image inpainting [2], etc.) is needed rather than general linear signal processing. Figure 1, Tables I and II show an example of recovery from Lena image with missing data by image inpainting technique. Furthermore, recovery techniques for images (videos) with missing data are akin to image and video processing techniques for old film sequences including film displacement, blotch and line scratch. Recovery techniques developed in Refs. [4] and [5] can be applied.