Gradual Divergence for Seamless Adaptation: A Novel Domain Incremental Learning Method