In this article, an excessive method of learning algorithm has been developed in consideration of the efficient detection of picture forgery for both copy and move as well as split picture forgery. The effective pre-processing, better transformation function extraction, and optimal boundary detection make it extremely efficient to detect the classification of the image. The input image is pre-processed by the median filter to increase the predictive precision of the classification in the first step of this proposed algorithm. The picture is then divided into blocks. To detect forgery, an effective boundary detection method active contour snake (ACS) is suggested. To extract the feature vector from the object boundary detection area, the Contourlet Transform (CT) is then inserted. The vector similarity is calculated with the distance bat algorithm (DBA). Finally, the authentic and tempered picture is labeled using hybrid extreme learning machine-Fuzzy Sigmoid Kernel (HELM-FSK) for both blind processes. Hybrid methods of image processing for efficient detection of image classification both for copy-move and splice images are given in this paper. HELM-FSK is used for forensic images of classified authentic images and classification. The forgery picture is classified by HELM-FSK, in which differential evolution (DE) is used to maximize the weight of HELM. The data set is qualified and checked and the findings are comparable with traditional approaches to check the classification precision.