TDMEC, a new measure for evaluating the image quality of color images acquired in vision systems


In robotic imaging systems, images are often subject to additive Gaussian noise and additive noise in the color components during image acquisition. These distortions can arise from poor illumination, excessive temperatures, or electronic circuit noise. Imaging sensors required to perform real time enhancement of images that is best suited to the human visual system often need parameter selection and optimization. This is achieved by using a quality metric for image enhancement. Most image quality assessment algorithms require parameter selection of their own to best assess the image quality. Some measures require a reference image to be used alongside the test image for comparison. In this article, we introduce a no-parameter no-reference metric that can determine the best visually pleasing image for human visual perception. Our proposed metric is algorithm independent such that it can be utilized for a variety of enhancement algorithms. Measure of enhancement methods can be categorized as either spatial or transform domain based measures. In this article, we present a DCT transform domain measure of enhancement to evaluate color images impacted by additive noise during image acquisition in robotics applications. Unlike the spatial domain measure of enhancement methods, our proposed measure is independent of image attributes and does not require parameter selection. The proposed measure is applicable to compressed and non-compressed images. This measure could be used as an enhancement metric for different image enhancement methods for both grayscale and the color images.

DOI: 10.1109/TePRA.2015.7219662

6 Figures and Tables

Cite this paper

@article{Samani2015TDMECAN, title={TDMEC, a new measure for evaluating the image quality of color images acquired in vision systems}, author={Arash Samani and Karen Panetta and Sos S. Agaian}, journal={2015 IEEE International Conference on Technologies for Practical Robot Applications (TePRA)}, year={2015}, pages={1-5} }