This paper introduces a novel image compression method that employs a predefined tree structure to encode each non-overlapping block of an image. Unlike the conventional Vector Quantization (VQ) approach, which maps image blocks to a fixed codebook, our method utilizes a hierarchical tree structure to represent image blocks, allowing for more efficient encoding and better preservation of visual details. The tree structure is designed by considering whether the image block’s content is smooth or complex; for complex content, the block is classified as one of five edge types. Extensive experimental results demonstrate that our proposed method achieves superior compression efficiency compared to VQ. Furthermore, the visual quality of the decompressed images is consistently higher, exhibiting fewer artifacts and better fidelity to the original image. This advancement in image compression technology offers significant improvements for applications requiring high-quality image storage and transmission under limited bandwidth constraints.