We proposed, a perceptual robust image hashing using random Gabor filtering and Markov Absorption Probability. The global as well as local features are extracted for the formation of the hash. Gabor filter is applied to extract global features. The conventional Gabor filter is modified to have good invariant property against rotation and the rotation-invariant filter is randomized to facilitate secure feature extraction. Markov Absorption Probability is applied for detection of salient regions and then position and texture vectors are calculated to extract the local features. Individual element saliency is obtained from Markov absorption probability. Mathematically, Markov absorption probability is determined by virtual boundary nodes, both left and top nodes, having maximum similarity. Secret keys are incorporated in feature extraction and hash construction for security. The use of Markov Absorption Probability improves the forgery classification. A test image subjected to content-preserving operation is considered for evaluation of the algorithm performance. A superior robustness is observed in the proposed algorithm comparative to the state-of-art algorithms, specifically in rotation performance.