In this paper, it is presented a detailed description of a proposal of an effective image encoder suitable for hardware realizations. A modelling block based on the blending prediction method and an encoding block performing a contextual, adaptive arithmetic encoding are proposed. The encoding operation in both lossless and near-lossless modes are described. It is analyzed how to lossless transform colours to obtain the best solution. An effectiveness analysis is also done by comparison of the compression rate of the proposed method with another hardware realization and a number of software implementations known from literature. Some suggestions are made regarding the implementation of the algorithm as the video sequence encoding system working in the real-time utilizing an efficient Network on Chip architecture.