Comparative study of color image enhancement techniques

Abstract

This paper presents a color image enhancement algorithm based on shifting the histogram of logarithmic DCT coefficients of the luminance component. The novelty of this work lies in utilizing both the spatial domain histogram concept and transform domain enhancement techniques to perform the enhancement. This paper also demonstrates a quantitative measurement based upon contrast entropy to choose the optimal parameters and determines the effectiveness of the method. In addition, different color spaces including HSV, YCbCr, and Principle Component Analysis (PCA) have been examined for the algorithm to determine the best visual results. We also present a comprehensive review of four commonly used color image enhancement techniques, such as mutiscale retinex with color restoration, multi contrast enhancement, multi contrast enhancement with dynamic compression and color image enhancement by scaling, and compare them with the proposed histogram shifting techniques. Computer simulations and analysis show that the histogram shifting method outperforms the commonly used methods for most images.

Spectrum Classification of Easily Confused Ground Objects in ALI Remote Sensing Image Based on Texture Features

“…Haralick, et al have proposed 14 kinds of texture statistics of gray level co-occurrence matrix [8][9] , this paper choose the 4 commonly used texture statistics that is Contrast, Dissimilarity, Mean and ASM to join the classification with the spectrum information [10][11] .…”

3-D Video Enhancemnt Based on Histogram Equalization and Edge Sharpening

“…The method used a multi-scale contrast measure in Laplacian Pyramids for Breast Cancer Detection, then enhanced image matches the Human Vision System (HVS) better and has better visual quality. In [13], Xia et al presents a color image enhancement algorithm based on shifting the histogram of logarithmic DCT coefficients of the luminance components.…”