Analyzing and preserving the authenticity and integrity of multimedia content is a major function of multimedia forensics. In this paper, our main objective is to analyze images that are forged with easily available advanced software tools. Forged images are created in a realistic way to harass and humiliate women via fake porn images, which is a major concern. There are image forensic tools that help the forensic investigator analyze forged images, but are not convincing in detecting the type of forgery and forged region. We propose a pre-trained hybrid LSTM-CNN-based model to detect the forged region with an improved SIFT algorithm. When compared with existing models, the proposed model provides good performance accuracy. The proposed model is beneficial for law enforcement agencies to apply the model for developing forensic tools and also helps forensic analysts investigate the forgery of pornographic images.