Intelligent Engineering Systems through Artificial Neural Networks, Volume 16
65 A New Image Quality Metric for Evolved Weighted Voronoi Image Segments
Download citation file:
- Ris (Zotero)
- Reference Manager
This paper introduces a new metric for evaluating the quality of images that have been processed by a segmentation algorithm which replaces the pixels in each segment with pixels of a single color. Segmentation algorithms divide an image into homogeneous regions; they are useful in image processing. The novel segmentation algorithm used here is an evolutionary algorithm that evolves weighted Voronoi tessellations. This algorithm optimizes zero-variance image segments that capture irregular feature boundaries more effectively than the traditional segmentation algorithms. Many robust metrics are available for regular-shaped blocks with non-zero variance in each block. The values returned by these metrics for zero-variance image segments are inconsistent with perceived image quality. To address these limitations, a new image quality metric, Decision Quality Index of Images (DQII) is proposed. DQII takes into consideration the spatial frequency change, the luminance distortion, and the color variance between the original image and the segmented image.