<p>The document images are considered as one of the trustworthy source of crucial information in various domains such as passports, driving licenses, banking statements, medical reports of patients, courts of law related documents, military documents, etc. But due to the advancement of technology, document images can be tampered easily by the malignant users using different image editing tools for their personal benefits. Hence, it is a serious research concern to check the authenticity of the document images. We propose a method for copy-move forgery detection in document images. We utilize contour-based superpixel segmentation technique to segment document images. For keypoint extraction from the segmented image, we utilize Haar-Difference of Gaussian (Haar-DoG) technique. The descriptor corresponding to keypoints of the selected segments are computed using Partial Intensity Invariant Feature Descriptor (PIIFD) technique. We identify the similar keypoint descriptors using Dynamic Time Warping (DTW) algorithm followed by coarse and fine matching. We obtain the cluster of keypoints using density and distance-based sampling technique. To perform matching of clusters of keypoints and outlier removal, we utilize Locality Preserving Matching (LPM) technique followed by the analysis of local geometrical relationships. Further, we apply Singular Value Decomposition (SVD) and Histogram of Oriented Gradients (HOG) techniques to compute feature vector corresponding to the image blocks of the localized copy-move pairs for distinguished identification of authentic and tampered regions of manipulated document images. For experimental analysis, we have utilized Copy-Move ID (CMID), Real Text Manipulation, CoMoFoD, and CASIA v2.0 datasets. Experimental outcomes show the effective robustness of the proposed scheme for detection of tampered document images even in the presence of several post-processing attacks.</p>