Abstract
Bilateral Filtering (BF) is an effective edge-preserving smoothing technique in image processing. However, an inherent problem of BF for image denoising is that it is challenging to differentiate image noise and details with the range kernel, thus often preserving both noise and edges in denoising. This letter proposes a novel Dual-Histogram BF (DHBF) method that exploits an edge-preserving noise-reduced guidance image to compute the range kernel, removing isolated noisy pixels for better denoising results. Furthermore, we approximate the spatial kernel using mean filtering based on column histogram construction to achieve constant-time filtering regardless of the kernel radius’ size and achieve better smoothing. Experimental results on multiple benchmark datasets for denoising show that the proposed DHBF outperforms other state-of-the-art BF methods.
Highlights
Edge preservation and smoothing of the content in an image is a basic problem in machine vision, computer graphics, and image processing
We extend the existing Bilateral Filtering (BF) frameworks by proposing a novel Dual-Histogram BF (DHBF) method for better denoising results
To keep the constant time computation of BF, Porikli implemented the integral histogram database, avoiding the collection of redundant data in the spatial domain [14]. He et al provided high accuracy of local window matching around color edges for image smoothing by injected locality-sensitive histogram to linear-time BF [15]
Summary
Edge preservation and smoothing of the content in an image is a basic problem in machine vision, computer graphics, and image processing. The salient regions of an image were extracted for image segmentation by the re-blurring model with bilateral and morphological filtering [3]. The joint bilateral filter was modified to up-sample the depth image with the high-resolution edge guidance for image super-resolution [4]. Licciardo et al implemented the edge-preserving image smoothing hardware architecture for real-time 60 fps tone mapping in 1920 × 1080 pixels of image [7]. The work [8] extracted fine details from underexposed or overexposed images using content-adaptive bilateral filtering in the gradient domain for multi-exposure image fusion. The bright-pass bilateral filtering was presented to estimate the scene illumination for low-light image enhancement [9]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
