Low Dynamic Range (LDR) Histogram Equalization using Haar Wavelet Transform (HWT)
Web link of the article is here.
This is a new model to enhance the image contrast using the Haar Wavelet Transform (HWT) on Histogram Equalization. This proposed model gives less brightness improvement, higher contrast enhancement, lower Mean Square Error rates.
All these experimental studies are conducted using a large list of performance metrics below:
1. MSE (Mean Squared Error)
2. PSNR (Peak Signal-to-Noise Ratio)
3. SSIM Structural Similarity Index)
4. CII (Contrast Improvement Index)
5. UIQ (Universal image quality index)
6. PCC (Pearson 2-D correlation coefficient)
7. PCQI (Patch-based Contrast Quality Index)
8. QRCM (Quality-aware Relative Contrast Measure)
9. Absolute Mean Brightness Error (AMBE)
10. Mutual Information (MI)
All these metrics are obviously for a detailed examinations.
Availability (MATLAB, Java Processing Codes and Test Datasets)
Original 30 test QSIC images, the implemented MATLAB codes that determines the image qualities between the original and generated pictures, Java Processing Codes that generates the test pictures for the proposed model, and the whole test results and related documents can be publicly reachable from this web site for evaluation. The Excel table contains all the experimental results in terms of the 10 metrics listed above.
All the MATLAB codes and the processed image outputs of the Histogram Equalization methods can be downloaded from the links below for evaluation and tests.
1. LDR-HE (Low Dynamic Range Histogram Equalization) with the all coefficient (alpha) parameters.
2. HISTEQ (Global Histogram Equalization or Conventional Histogram Equalization)
3. CLAHE (Contrast Limited Adaptive Histogram Equalization)
4. LHE (Local histogram equalization)
5. MMSICHE (Median-Mean Based Sub-Image-Clipped Histogram Equalization)
6. BPDFHE (Brightness preserving dynamic fuzzy histogram equalization)
7. EHS (Exact histogram specification)
8. ESIHE (Exposure based Sub-Image Histogram Equalization)
9. BBHE (Mean preserving Bi-Histogram Equalization)
10. MMBEBHE (Minimum Mean Brightness Error Bi-Histogram Equalization (MMBEBHE)
11. DSIHE (Dualistic Sub-Image Histogram Equalization (DSIHE)
12. RMSHE (Recursive Mean-Separate Histogram Equalization)
13. RSIHE (Recursive Sub-Image Histogram Equalization)
14. RSWHE (Recursive Separated and Weighted Histogram Equalization)
15. NMHE (Non-parametric modified histogram equalization)
There is an ILLUSTRATION of LDR-HE in order to cover all the details.
The document is here.
The LDR-HE Outputs for RGB images according to the Coeff. parameters can be listed as:
In the first line, the values are the mean values over the dataset including 30 test images. At the second line, the values are the standard deviation values of the corresponding measurements.
The LDR-HE Outputs for Gray Scale images according to the Coeff. parameters can be listed as:
The Full Comparative Outputs of the methods for RGB images can be listed as:
The Full Comparative Outputs of the methods for Gray Scale images can be listed as:
For any question, do not hesitate to contact with me:
farukbulut(at,the special sign)arel(dot)edu(tr)
Please don't forger to refer this study as:
Faruk Bulut, (2021), Low dynamic range histogram equalization (LDR‑HE) via quantized Haar wavelet transform, The Visual Computer,https://doi.org/10.1007/s00371-021-02281-5