CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
WEIGHT MODEL ANALYSIS FOR IMAGE COMPRESSION BASED ON WAVELET TRANSFORM
 |
Read | Lyasheva Stella A., Morozov Oleg G., Shleymovich Mikhail P. WEIGHT MODEL ANALYSIS FOR IMAGE COMPRESSION BASED ON WAVELET TRANSFORM // Caspian journal : management and high technologies. — 2020. — №3. — pp. 9-22. |
Lyasheva Stella A. - Kazan National Research Technical University named after A.N. Tupolev-KAI, salyasheva@kai.ru
Morozov Oleg G. - Kazan National Research Technical University named after A.N. Tupolev-KAI, ogmorozov@kai.ru
Shleymovich Mikhail P. - Kazan National Research Technical University named after A.N. Tupolev-KAI, mpshleymovich@kai.ru
A new method of image compression based on multi - resolution wavelet transform is proposed. As a result of its execution, data is generated that contains information about image sizes, the initial transform level, approximating coefficients, maps of the significance of detail coefficients, and significant detail coefficients. Maps of the significance contain binary values that determine the need for corresponding detailed coefficients for image recovery. The algorithm for constructing significance maps contains the following steps: multi - resolution wavelet transformation, estimation of the difference energy value in each pixel at all levels of the multiscale representation using the detailing coefficients, calculation of weight values for the detailing coefficients at each level in terms of their contribution to image perception, taking into account the relationship between the scale levels, and threshold weight processing. As a significant detailing coefficient at every level are selected, the weight of which is greater than a predetermined threshold value. The proposed method reduces the image size by discarding insignificant detail coefficients. To increase the compression efficiency, entropy encoding of elements of significance maps and significant detailing coefficients is performed using the adaptive Huffman method. The proposed method allows you to obtain compression and recovery characteristics that are comparable and higher in quality for test images than the corresponding characteristics of popular image presentation formats. To reduce the time spent on image analysis and decision - making, it provides the possibility of parallel implementation and performing progressive compression and recovery.
Key words: image processing, image analysis, image wavelet transform, image weight model, image compression