Population genetic models play a significant role in human genetic research, since they promise to provide a better understanding of both evolution of normal variations of the genomes as well as development of disease promoting genomic segments. In this paper, we propose and computationally simulate a model of evolution for unique and segmentally duplicated regions of human genome, which manifest in genome-wide Variations in Copy Number. Copy Number Variations (CNVs) are known for their possible association with genomic diseases and predisposition to various health conditions, and yet, the underlying population genetic models, needed for association studies, have remained largely undeveloped. Additionally, Segmentally Duplicated (SD) regions of the human genome are of a specific interest, since they exhibit a complex behavior of copy number changes and are specifically known for catalyzing pathogenic and unstable genomic rearrangements. Our main contribution is in suggesting mechanisms for evolution of Copy Number changes in regions of segmental duplication of the human genome, which is essential for association studies involving CNV markers and estimation of parameters of stochastic diffusion that determine asymptotic behavior of such evolutionary processes.