An efficient psychovisual threshold technique in image compression

• Main Author: Ferda, Ernawan
• Format: Thesis
• Language: English; English
• Published: 2014
• Subjects: T Technology (General); TA Engineering (General). Civil engineering (General)
• Online Access: http://eprints.utem.edu.my/id/eprint/14899/1/AN%20EFFICIENT%20PSYCHOVISUAL%20THRESHOLD%20TECHNIQUE%2024pages.pdf; http://eprints.utem.edu.my/id/eprint/14899/2/An%20efficient%20psychovisual%20threshold%20technique%20in%20image%20compression.pdf; http://eprints.utem.edu.my/id/eprint/14899/; https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=92062
• Institution: Universiti Teknikal Malaysia Melaka

Nowadays, psychovisual model plays a critical role in an image compression system. The psychovisual threshold gives visual tolerance to the human visual system by reducing the amount of frequency image signals. The sensitivity of the human eye can be fully explored and exploited in the qualitative experiment by describing what has been seen or by image quality judgment. However, the result of the psychovisual threshold through qualitative experiment depends on the test condition of the human visual systems and through repetitive viewing sessions. In a modern image compression, there is a need to provide some flexibility to obtain quality levels of the image output based on user preferences. The concept of psychovisual threshold is designed to determine quality levels of the image output. The psychovisual threshold represents an optimal amount of frequency image signals in image compression. This research proposes the psychovisual threshold through a quantitative experiment that can automatically predict an optimal balance between image quality and compression rate in image compression. The contribution of its frequency image signals to the image reconstruction will be the primitive of psychovisual threshold in image compression. It is very challenging to develop a psychovisual threshold from the contribution of the frequency image signals for each frequency order. In this research, the psychovisual threshold prescribes the quantization values and bit allocation for image compression. The psychovisual threshold is the basic primitive prior to generating quantization tables in image compression. The psychovisual threshold allows a developer to design adaptively customized quantization values according to his or her target image quality. The psychovisual threshold is also elementary and primitive for generating a set of bit allocation for frequency image signals. A set of bit allocation based on psychovisual threshold assigns the amount of bits for frequency image signals. A set of bit allocation refers to the psychovisual threshold instead of the quantization process in image compression. This research investigates the basic understanding of the psychovisual threshold in image compression. The experimental results provide significant improvement in the image compression. The psychovisual threshold which is presented as quantization tables, customized quantization tables and as a set of bit allocation gives a significant improvement on both of the quality of the image reconstruction and the average bit length of Huffman code. This research shows that psychovisual threshold is practically the best measure for optimal frequency image signals on image compression.