Adaptive directional wavelet transform based on directional prefiltering.

Area of Science:

• Image Processing
• Signal Processing
• Computer Vision

Background:

• Adaptive directional wavelet transform (WT) offers superior image coding performance by analyzing diagonal orientations alongside traditional horizontal and vertical directions.
• Conventional adaptive directional WT methods often incur high computational complexity, limiting their practical application.
• Efficient computation of transform directions is crucial for balancing performance and speed in image processing.

Purpose of the Study:

• To develop computationally efficient methods for determining optimal transform directions in adaptive directional WT.
• To maintain or improve image coding performance while reducing the complexity of adaptive directional WT.
• To explore the utility of transform direction data for content-based image retrieval applications.

Main Methods:

• Proposed two novel prefiltering techniques for efficient calculation of adaptive directional WT.
• Method 1: Prefiltering using a 2-D filter bank.
• Method 2: Prefiltering using a 1-D directional WT along two fixed directions.

Main Results:

• The proposed direction calculation methods achieve image coding performance comparable to conventional approaches.
• Significant reduction in computational complexity compared to existing methods.
• Transform direction data generated by the proposed methods demonstrates potential for enhancing content-based image retrieval accuracy.

Conclusions:

• Efficient prefiltering strategies can substantially decrease the computational cost of adaptive directional WT.
• The proposed methods provide a practical solution for high-performance image coding with reduced complexity.
• The derived transform direction information offers a valuable feature for content-based image retrieval systems.