Correlation Properties of Color Histograms in the Case of Image Quality Decreasing
DOI:
https://doi.org/10.34185/1562-9945-6-155-2024-08Keywords:
distortion of images, color histogram, HS* color spaces, Pearson correlation, OpenCV.Abstract
In image recognition tasks, objects are identified by examining various features such as texture, color, contour detection, and statistical or semantic descriptions. One widely used approach for extracting image attributes is the analysis of intensity histograms. While the traditional RGB color model is commonly used in digital image processing, it is often more effective to analyze color properties in HS* systems (such as HSL, HSV, and HSI) since these systems more closely resemble the spectral representation of color. A key characteristic shared by these three systems is the use of the H (Hue) coordinate, which is represented as an angular value within a cylindrical coordinate system. The paper investigates the possibility of using color histograms generated in HS* spaces for identifying images that have undergone various types of distortions. The CQ100: A High-Quality Image Dataset for Color Quantiza-tion Research was chosen for the research. The non-quantized section of the CQ100 dataset consists of 100 RGB images in PNG format, each with a resolution of 768×512 pixels and a color depth of 24 bits. The study examines how different distortions, which can occur during real-time photo and video capture, affect the color properties of images. Specifically, the re-search focuses on distortions caused by rotation, noise, blurring, and optical aberration. His-tograms were compared using the Pearson cross-correlation coefficient, and the findings re-veal that the correlation remains high for the same image despite the applied distortions. Conversely, the correlation coefficient between different images is low for most of the studied objects. These results suggest that color histograms could be effectively used for image identi-fication tasks, even when images are significantly distorted, as is common in image registra-tion processes. The applicability of correlation detection as a method for histograms com-parison is considered regardless of the relative simplicity of its calculation. This approach could contribute to the development of faster image recognition systems.
References
Burger W., Burge M.J. Principles of Digital image processing: Core Algorithms // Springer. – 2009. – 344 P.
Gonzalez R.C., Woods R.E. Digital Image Processing // [4th ed.] – Pearson Education Ltd. – 2017. – 1192 P.
Duda R.O., Hart P.E., Stock D.G. Pattern Classification // [2nd ed.] – Wiley. – 2000. – 688 P.
Agoston M.K. Computer Graphics and Geometric Modeling: Implementation and Algorithms // [5th ed.] – Springer. – 2005. – 928 P.
Celebi M.E., Pérez-Delgado M.-L. CQ100: A High-Quality Image Dataset for Color Quantization Research // Mendeley Data, V2. – 2022. DOI: 10.17632/vw5ys9hfxw.2
Celebi M.E. Forty years of color quantization: a modern, algorithmic survey // Artif. Intell. Rev. –2023. – Vol. 56. – pp. 13953–14034. DOI: 10.1007/s10462-023-10406-6.
Bland M. An Introduction to Medical Statistics // [4th ed.] – Oxford University Press. – 2015. – 446 P.
OpenCV Documentation. Basic Operations on Images. URL: https://docs.opencv.org/4.x/d3/df2/tutorial_py_basic_ops.html (дата звернення/Accessed: 15.08.2024).
Furgala Yu., Velhosh A., Velhosh S., Rusyn B. Using Color Histograms for Shrunk Images Comparison // 2021 IEEE XIIth International Conference on Electronics and Information Technologies (ELIT). – 2021. – pp. 130-133. DOI: 10.1109/ELIT53502.2021.9501117
Miller S., Childers D. Probability and Random Processes With Applications to Signal Processing and Communications // [2nd ed.] – Academic Press. – 2012. – 522 P.
OpenCV Documentation. Smoothing Images. URL: https://docs.opencv.org/4.x/d4/d13/tutorial_py_filtering.html (дата звернення/Accessed: 15.08.2024).
Kingslake R., Johnson R.B. Lens Design Fundamentals // [2nd ed.] – Academic Press. – 2009. – 576 P.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 System technologies
This work is licensed under a Creative Commons Attribution 4.0 International License.
