In this article, we propose a novel image dependent based Reversible Data Hiding (RDH) scheme. Suitable regions for data hiding within an image is identified in spatial domain as well as in Integer Wavelet Transform (IWT) domain. At first, the cover image is split into low and high frequency image blocks based on eigen value based R-value measure of a particular block. Based on the classification, these image blocks arerearranged resulting to a scrambled image. We propose a novel algorithm to generate a permutation key (K₁)from scrambled image block sequence. First L blocks (key K₂) of this scrambled image correspond to low frequency image blocks whereas remaining blocks corresponds to high frequency. This key generation algorithm enhances the security as well as eliminates the need of embedding a significant bookkeeping information as metadata, improving the payload. Further, data hiding is performed in low frequency image blocks using Difference Expansion (DE) method. Remaining high frequency blocks are decomposed by using IWT. Further, DE is employed on the resultant transformed coefficients leading to increase in payload. Proposed scheme provides enhanced security due to the generation of permutation keys. We extensively evaluate the proposed method on standard image database. Results show that our proposed method provide better ratedistortion performance than the state-of-the-art methods.