Keynote address: Exciting moments of research: Revisiting a few problems on image processing

In this talk, I would like to revisit some of the research problems in the area of image processing, on which I worked in the past and share my excitement of conceiving those ideas and unfolding them to their detailed exposition. In my talk, I will briefly cover different research problems on topics, such as geometric analysis of images, fractal image and video compression, and color image processing. A brief outline of each of these topics is provided below. There are various applications of digital geometry in processing images, specifically bi-level or binary images. Efficient preprocessing techniques such as, thinning, segmentation, wire-framing etc. are developed exploiting the geometry in the discrete space. There exist also various digital metrics such as octagonal metrics, which are found to be useful in approximating Euclidean metric. Moreover, Medial Axis Transform (MAT) of images defined in those metric spaces, are efficient to compute and process for various purposes, such as applying geometric transformation to the objects, computation of cross-sections of 3-D objects, computation of surface normals, etc. Fractal models of images fascinated many researchers and in the nineties (of the last century), there were many efforts for developing efficient compression and decompression algorithms using fractal representation. One of our interesting work in this regard was to study the convergence of the decompression algorithm, which also helped developing efficient linear time decoding algorithm. The concept was also found to be useful enough for its application in video compression. In color processing, we would address various issues on deomosaicing, constancy and enhancement. Some of these algorithms were also extended in the compressed domain.