Secure Image Retrieval of Poor Quality Images by Combining LE-GAN, Arnold Mapping and Logistic Mapping

Abstract

The quantity of image data is increasing rapidly with the discovery of big data and internet technology. Currently, the majority of image retrieval techniques rely on plain text images. It is a threat to several professional fields like medicine, the military, and the government. One of the limitations of the current data model is that it is difficult to effectively retrieve images with low quality samples. LE-GAN networks can be utilized to enhance the appearance of images. Then the enhanced images are fed into the network for retrieving images securely. Using a deep artificial neural network model to extract characteristics from training data can increase the security of an image's network transmission. Then, image retrieval [1] is devised and coupled with an image encryption technique that complements and secures image retrieval [1]. The recommended method can comfy the ciphertext images' retrieval and also can increase retrieval performance. Feature extraction has accomplished the usage of AlexNet and a chaotic algorithm is used as an encryption algorithm. To safeguard the image feature facts, the encryption technique is split into components so that the image information can nevertheless be successfully covered. To enforce the feature of image encryption, Arnold Mapping, and 2D Logistic Mapping are employed.