IMPLEMENTATION OF DIGITAL IMAGE METAMORPHOSIS

Abstract

Image Metamorphosis has received lot of interest in the recent years, much of it due to its large range of applications, mainly, in the fields of Remote Sensing, Medical Imaging, Computer Animation and Visual Effects. In Satellite Imaging, distortions are generally caused by surface curvature and oblique viewing angles. In ultrasonic Medical Imaging, distortions are caused by the varying speed of sound in different media. These distortions must be rectified to obtain the correct coordinates and correct image. However, in the field of Computer animation, the goal is not a geometric correction, but inducing a geometric distortion. These geometric corrections or distortions are performed by applying a Geometric Transformation, A Geometric Transformation is an operation that redefines the spatial relationship between points in an image. The term " IMAGE METAMORPHOSIS" describes methods for deforming images to arbitrary shapes by applying different geometric transformations. These geometric transformations are generally implemented by applying Scanline algorithms which perform the operation in two or more passes to overcome the problems of forward mapping. Even though they exploit simplifications in mathematical calculations, memory access and take less time, they are inefficient for applications where space dominates speed, because they require large amount of temporary storage at run time. An approach called "Feature-Based Metamorphosis" has been implemented using inverse mapping to overcome this problem. In this work the metamorphosis has been divided into two categories depending on the complexity of the transformation. The First one includes simple transformations like rotation, scaling etc. The second one includes complex transformations like warping and morphing. Simple transformations for rotation and scaling have been implemented and a comparison between the Scanline algorithms and Feature-based approach has been done in terms of time and space requirements and their advantages and disadvantages have been examined. Much of the concentration has been given to Morphing and its application in the fields of visual effects. Three important morphing techniques, Two Pass Mesh Warping which is a Scanline algorithm, Feature-Based Morphing, and Scheduled Fourier Morphing have been implemented and a detailed comparative study is drawn. Advantages & disadvantages of each have been examined. Several examples are illustrated with resulting images to explain each method and their application in different fields.