Linear models of cumulative distribution function for content-based medical image retrieval.

Area of Science:

• Computer Science
• Image Processing
• Pattern Recognition

Background:

• Efficient image retrieval is crucial in large databases.
• Existing image matching techniques can be computationally intensive, leading to long retrieval times.

Purpose of the Study:

• To propose a novel image matching technique for reduced retrieval time.
• To introduce and evaluate bit plane histogram and hierarchical bit plane histogram approaches.
• To compare the proposed Cumulative Distribution Function (CDF) based technique with existing methods.

Main Methods:

• Approximating the CDF of query and database images using piecewise linear models.
• Utilizing slope and intercept parameters at various grayscale intervals for CDF approximation.
• Comparing images based on estimated slopes and intercepts of their CDFs.
• Employing contiguous lines to represent CDFs for efficient comparison.

Main Results:

• The proposed CDF-based image matching technique offers a considerable reduction in retrieval time.
• Approximation of CDFs with lines reduces feature dimensions, enhancing matching speed.
• The method allows for the comparison of images with different sizes due to the dynamic range of CDF (0 to 1).

Conclusions:

• The CDF-based image matching technique provides a significant improvement in retrieval speed.
• The use of piecewise linear models for CDF approximation is effective for image matching.
• This approach offers a scalable and efficient solution for large-scale image retrieval.