XGB-PV: A Radical Approach to Non-linear Image Steganography Using Ensemble Learning for Validation and Extensive Comparative Analysis

Billion people use the Internet everyday. Web-based and independent apps need internet access to work. In order to ensure the security of data communicated over the Internet at all times, cryptography, encryption/decryption and data concealment methods have been created. Instead of security protocols, hackers might readily access the secret data due to the lack of protection. Steganography conceals a message in an image, audio file, video or text such that no one suspects it. Images may be steganographically altered to conceal communications by carefully designing a cover picture. This paper proposes a new non-linear approach for secret text embedding in image and also compares edge-based and non-edge based embedding performance. In addition, an eXtreme Gradient Boost Prediction and Validation (XGB-PV) model is created to evaluate the suggested technique’s validity in relevant metrics achieving highest R2 score of 99.99% and model score of 96.24%.