It is essential to maintain a proper balance between the embedding capacity, imperceptibility and robustness in image-based reversible data hiding (RDH) applications. Most of the spatial-domain RDH techniques focus more on embedding capacity and imperceptibility as compared to robustness, whereas transform-domain techniques concentrate mainly on imperceptibility and robustness. Hence, there is a need to implement a technique which maintains a proper trade-off among all the three parameters. A scheme that uses a combination of spatial- and transform-domain (hybrid) techniques is proposed here. A threshold-based difference histogram modification (TDHM) is used as a building block for data embedding and extraction. In the proposed technique, the cover image is systematically modified using spatial-domain rhombus prediction-based prediction error histogram modification and integer Haar wavelet transform (IHWT)-based TDHM (IHWT-TDHM) techniques. The experimental results show that the stated methodology accomplishes a higher embedding capacity while maintaining good imperceptibility and robustness compared to that of the direct spatial or transform RDH techniques. © 2018, Springer Science+Business Media, LLC, part of Springer Nature.