Effective Guided Image Filtering for Contrast Enhancement

Zongwei Lu,Kang Li,Bangyuan Long,Fajin Lu

doi:10.1109/lsp.2018.2867896

Zongwei Lu, Kang Li + Show 2 more

Open Access

https://doi.org/10.1109/lsp.2018.2867896

Copy DOI

Abstract

Although the guided image filtering (GIF) has an excellent edge-preserving property, it is prone to suffer from the halo artifacts near the edges. Weighted GIF and gradient-domain GIF try to address the problem by incorporating an edge-aware weighting into GIF. However, they are very sensitive to the regularization parameter and the halo artifacts will become serious as the regularization parameter increases. Moreover, noise in the background is often amplified because of the fixed amplification factor for the detail layer. In this letter, an effective GIF is proposed for better contrast enhancement. First, the average of local variances for all pixels is incorporated into the cost function of GIF for preserving the edges accurately in the base layer. Second, the amplification factor for the detail layer is calculated in a content-adaptive way for suppressing the noise while boosting the fine details. Experimental results show that the proposed filter is more robust to the regularization parameter and can produce visually pleasing output images. Compared to GIF and its related filters, halo artifacts and noise are reduced or attenuated by the proposed filter significantly.

Highlights

T HE edge-preserving smoothing techniques have many applications in image processing and computer vision, including contrast enhancement [1], tone mapping of high dynamic range images [2], structure extraction from texture [3], fusion of differently exposed images [4], and single image haze removal [5]
Considering the local variances of the windows for the filter size, gradient-domain GIF (GGIF) defines edge-aware weighting as follows [15]: ΓG (p p=1 where χ(p ) = σG,1 (p )σG,ζ1 (p ), σG2,ζ1 (p ) is the variance of I in the window ΩG,ξ1 (p ), and ξ1 is the window size of the filter. While both weighted GIF (WGIF) and GGIF can preserve the edges more accurately than guided image filtering (GIF) theoretically, experimental results show that the halo artifacts cannot be avoided for strong edges when they are applied for contrast enhancement
In order to compare the performance of edge-preserving for GIF, WGIF, GGIF, and effective GIF (EGIF) fairly, the amplification factor for the detail layer is set as 5 for EGIF

Summary

INTRODUCTION

T HE edge-preserving smoothing techniques have many applications in image processing and computer vision, including contrast enhancement [1], tone mapping of high dynamic range images [2], structure extraction from texture [3], fusion of differently exposed images [4], and single image haze removal [5]. Edge-preserving smoothing techniques can be categorized into two groups: global and local filters. Because the local filters cannot preserve the edges well as global filters, halo artifacts are prone to be produced near the edges in the resultant images, which is usually not favored for image processing and computer vision. Guided image filtering (GIF) is a well-known local filter for its excellent edge-preserving property and high efficiency [11]. Because the regularization parameter in the GIF is fixed, halo artifacts will be inevitably produced for strong edges in the output images. In order to reduce or attenuate the halo artifacts, weighted GIF (WGIF) [14] calculates the local variances of 3 × 3 windows of all pixels as the edge-aware weighting and preserves the edges better than GIF.

GIF AND RELATED WORKS

New Weighting

Content-Adaptive Amplification Factor

EXPERIMENTAL RESULTS

CONCLUSION