459 International Journal of Communication Networks and Information Security (IJCNIS) Vol. 13, No. 3, December 2021 Improved Deep Hiding/Extraction Algorithm to Enhance the Payload Capacity and Security Level of Hidden Information Marwa Ahmad 1 , Nameer N. El-Emam 2 and Ali F. AL-Azawi 3 1,2,3 Department of Computer Science, Philadelphia University, Jordan Abstract: Steganography algorithms have become a significant technique for preventing illegal users from obtaining secret data. In this paper, a deep hiding/extraction algorithm has been improved (IDHEA) to hide a secret message in color images. The proposed algorithm has been applied to enhance the payload capacity and reduce the time complexity. Modified LSB (MLSB) is based on disseminating secret data randomly on a Cover-Image and has been proposed to replace a number of bits per byte (Nbpb), up to 4 bits, to increase payload capacity and make it difficult to access the hiding data. The number of levels of the IDHEA algorithm has been specified randomly; each level uses a color image, and from one level to the next, the image size is expanded, where this algorithm starts with a small size of a Cover-Image and increases the size of the image gradually or suddenly at the next level, according to an enlargement ratio. Lossless image compression based on the run-length encoding algorithm and Gzip has been applied to enable the size of the data that is hiding at the next level, and data encryption using the Advanced Encryption Standard algorithm (AES) has been introduced at each level to enhance the security level. Thus, the effectiveness of the proposed IDHEA algorithm has been measured at the last level, and the performance of the proposed hiding algorithm has been checked by many statistical and visual measures in terms of the embedding capacity and imperceptibility. Comparisons between the proposed approach and previous work have been implemented; it appears that the intended approach is better than the previously modified LSB algorithms, and it works against visual and statistical attacks with excellent performance achieved by using the detection error (PE). Furthermore, the results confirmed that the Stego-Image with high imperceptibility has reached even a payload capacity that is large and replaces twelve bits per pixel (12-bpp). Moreover, testing is confirmed in that the proposed algorithm can embed secret data efficiently with better visual quality. Keywords: Steganography; Multi-level steganography; Deep hiding/extraction; Least significant bit; High payload; high security. 1. Introduction Currently, data hiding is one of the most important requirements, where data is sent and received online at every moment, and it requires security when it is shared over the Internet. Hiding information is an active research area, where confidential information is included in a carrier, such as photos and videos, to hide their presence while maintaining their visual quality. Researchers have provided different techniques for information concealment since the previous decade; they have focused on the load capacity and image quality Al-Shatanawi and El-Emam [1], El-Emam and Al-Zubidy [2], and Muhammad et al., [3]. In general, five main objectives are used to evaluate the performance of data-hiding algorithms, which include the embedding capacity, imperceptibility, security, robustness and complexity Darabkh et al. [4]. Recently, many analytical techniques have been developed to extract significant hidden information from Stego-Image s. Therefore, to avoid data extraction, several new steganography algorithms were improved by many researchers to make it difficult for the human visual system (HVS) to observe the difference between the stego- and Cover-Image s. In addition, to increase the security level and payload capacity, a multi-level steganography technique (MLS) was proposed by Sikarwar [5]; see Figure 1. Figure 1. Modes of Steganography The rest of the paper is structured as follows: In section 2, related works, and the proposed steganography algorithm specifications has appeared in section 3. In section 4 the requirements of improved deep hiding/extracting algorithm with the algorithm design and its implementation are presented in section 4. In section 5. The experimental results are discussed in. Finally, in section 6 the main conclusions of the proposed algorithm have been discussed. 2. Related Work Steganography is necessary due to the exponential growth and secret information of possible computer users over the Internet, so it is a technique of invisible information to keep secret data text, audio, and video files inside other data. Steganalysis is the technology that attempts to identify and extract the hidden data. Siswanto1 et. al [6] discussed that the performance of the encryption algorithm is low when considering the encryption speed. On the other hand, a simple encryption algorithm will have faster encryption speed; however, it generally provides low-security protection. Different techniques that use multiple bit replacement per pixel were discussed by El-Emam [7], El-Emam, and Al- Diabat [8], and El-Emam and Qaddoum [9]. These techniques are based on a modified least significant bit (MLSB) algorithm to work against the steganalysis process. The MLSB methods are precipitated by the notion that it is possible to detect the hidden data by using statistical analysis, such as the chi-square test or K-S test Ker [10]. Latika and Gulati [11] explained the types of nonlinear media that are used in data hiding techniques and classified these types of media into four main classes: text, image, audio, and