Reversible Data Hiding in Encrypted Images (RDHEI) is a technique that enables the embedding of additional information into encrypted carrier images, facilitating data extraction and exact restoration of the original image upon decryption. In this paper, enhancements to an RDHEI algorithm utilizing a block-wise pixel value prediction scheme have been analyzed and proposed. To better exploit spatial pixel correlations, seven additional prediction models have been introduced, along with the identification of reference pixels to improve the variable block-wise reconstruction of the carrier data. Furthermore, the integration of Huffman coding within the RDHEI scheme has been evaluated and compared with Extended Run-Length Encoding. Benchmark results against other RDHEI methods from the literature are presented at the end of the paper.