Identifying Forged Digital Image Using Adaptive Over Segmentation and Feature Point

Abstract

Wide availability of free image editing tools has made it very convenient to edit digital images. Image editing with wrong intensions is becoming a serious problem. The existing detection method deals with feature point equating and adaptive over segmentation in this work. Suggested conspire coordinates reality-based forgery and block-based detection techniques. Image tempering with bad intentions is becoming more common in the news in media field, patient data in medical field, and several other fields. This paper proposes an adaptive over segmentation-based detection of forged images. To begin, suggested algorithm slices the input picture into well separated and asymmetrical blocks suitably. At that point, the points extricated from scale invariant feature transform (SIFT) are again extricated from individual block. The features such obtained are called block features (BFs). To find labeled feature points, these BFs are equated with one another; this technique roughly shows the speculated fraud locales called surmised forgery locales. To identify the forgery locales correctly, the paper suggest an algorithm called forgery local extrication. Using the algorithm, the super pixels are replaced by the feature points and afterward combines similar local color features into the feature blocks to produce the unified locales. Lastly, to produce the detected forgery locales, we apply the morphological operation to the unified locales. The analyzed outcome and comparative analysis demonstrates that the proposed detection plan accomplishes prominent detection outcomes.