Spider Monkey Optimization-Based Image Data Forgery Detection Over Vehicular Cloud Computing

Images are commonly shared through various mediums such as newspapers, magazines, Internet, and scientific journals. But with the accessibility of software like Photoshop, GIMP, and Coral Draw, it has become increasingly difficult to differentiate between an original and a forged image. Traditional methods for detecting image forgery rely on handcrafted features, but they are limited in their ability to identify specific types of tampering based on certain features in the image. To overcome this obstacle, deep learning methods have been employed. Increased vehicular networking image data faces this challenge. The proposed image forgery detection uses hybrid model of convolution neural networks (CNN), spider monkey optimization technique (SMO), and support vector machines (SVMs). CNN extracts deep semantic features of images transferred as vehicular data over the vehicular cloud using error level analysis (ELA) techniques. SMO chooses the optimal features from deep semantic features. Binary classifications based on SVM provide whether the image is original or forged. The results demonstrate that the proposed method achieves a recognition accuracy of 96%, which is significantly higher than the same model without SMO, which achieved an accuracy of 93%.