Mid Sweden University

Encoding of image sequences is today a standard task for computers and home entertainment systems. The input material is often of shifting quality, i.e. corrupted by noise in various ways. In order to store the information in an effective way some kind of compression has to be applied. When video compression is applied on an image sequence, the result is sub optimal due to the introduced noise variations. This paper presents a new cubic median filter suitable for pre-processing of noise contaminated video sequences. The presented approach reduces the bit usage for MPEG-2 encoding with up to 62% for impulse noise impaired sequences and 38% for the corresponding Gaussian.

This paper presents a generalized architecture for the synthesis of application specific memory architectures for real-time image processing systems. The memory generation presented in this paper can handle both spatial and spatio-temporal memory models. The results show that the architecture efficiently solves the problems related to memory accesses for most of the available video image processing filters available at present.

This paper presents a comparison of local and global memory allocation for implementation of real-time video processing systems on FPGAs. The paper compares new methods to perform memory allocation using single and dual port block RAMs to memory allocation methods found in the literature. This investigation shows that the use of global allocation can reduce the memory size by up to 75 % for the two presented video processing systems.