My research interests include lossless data compression, image compression, data encryption, image and video codecs, parallel computing, CUDA, computer graphics.
Data compression is a challenging process with important practical applications. Specialized techniques for lossy and lossless data compression have been the subject of numerous investigations during last several decades. The purpose of a data compression process is to reduce the file size while minimizing the deterioration of information on the data. Reducing redundant data in a file is one of the common approaches used by most compression algorithms. There are two main types in data compression: lossy and lossless. In lossy compression, some information of the data may be ignored in encoding. When the encoded data is decoded, the original file and the decompressed file may differ. When portions of the data are not completely essential and minor loss in data is acceptable, lossy compression algorithms may be preferred in order to maximize compression gain. In lossless compression, while the quality of a file is preserved, the reduction in the file size may not be as good as in lossy compression.
Lossless Compression of Images
Pseudo-distance Transform (PDT) is used in lossless compression of color-mapped images. PDT uses neighboring pixels to create a pseudo-distance table and transforms index values into pseudo-distance table values, as shown below.
Concurrent Encryption and Lossless Compression using Inversion Ranks
For secure and efficient transmission or storage, data files are commonly compressed and encrypted. In this work, we introduce a cost-effective encryption method of files as a built-in component of a lossless compression algorithm, thus avoiding the added cost of employing two separate processes. During the compression process, we encrypt the frequency vector of the Inversion Ranking transformation and transmit it along with the compressed data. Since the frequency vector is required for decompression, no further encryption is necessary to secure the compressed file. Thus, encrypting only a relatively small section of data (1024 bytes) containing the frequency vector instead of the entire compressed file results in a substantial reduction in computational cost. We show in this study that the proposed concurrent encryption and lossless data compression technique is effective and resistant to common attacks using various cryptanalysis techniques on image and audio data sets. Datasets and executables