Complex Morphological Filtering For Serial, Parallel, GPU, SoC, PetaLinux And FPGA Execution

Abstract

Image processing is a vast field of research and extremely important for a large number of applications such as security systems, geoprocessing, medical technologies, etc. There are some applications that require a higher level of processing, requiring higher computing power. As an example of this requirement in image processing, morphological filtering with filter chains involving erosion and dilation may be used. This study aims to discuss the advantages and disadvantages between running these chains in a software context, using a personal computer, and in a hardware context, using the ZedBoard board running these filters in baremetal, FPGA and embedded linux modes. In addition, a discussion on the possibilities of parallel hardware processing, inspired by the multi-core environment and its power, will be discussed.