Quality estimation based multi-focus image fusion

Abstract

In this work, a fast Bilateral Solver for Quality Estimation Based multi-focus Image Fusion method (BS-QEBIF) is proposed. The all-in-focus image is generated by pixel-wise summing up the multi-focus source images with their focus-levels maps as weights. Since the visual quality of an image patch is highly correlated with its focus level, the focus-level maps are preliminarily obtained based on visual quality scores, as preestimations. However, the pre-estimations are not ideal. Thus the fast bilateral solver is then adopted to smooth the pre-estimations, and edges in the multi-focus source images can be preserved simultaneously. The edge-preserving smoothed results are utilized as final focus-level maps. Moreover, this work provides a confidence-map solution for the unstable fusion in the focus-levelchanged boundary regions. Experiments were conducted on 25 pairs of source images. The proposed BS-QEBIF outperforms the other 13 fusion methods objectively and subjectively. The all-in-focus image produced by the proposed method can well maintain the details in the multi-focus source images and does not suffer from any residual errors. Experimental results show that BS-QEBIF can handle the focus-level-changed boundary regions without any blocking, ringing and blurring artifacts.Index Terms-Multi-focus image fusion, confidence map, visual quality, the fast bilateral solver.

Guided adaptive interpolation filter

“…In addition to edge‐aware smoothing, these filters are broadly utilised in numerous applications in image processing and computational photography. Examples include image de‐noising [12, 17], detail enhancement [3, 18], image fusion [19, 20], texture smoothing [11, 12, 21–23], single image haze removal [24], tone mapping of high dynamic range (HDR) images [2, 3, 10, 23, 25], anomaly detection in hyper‐spectral images [26], object classification accuracy enhancement in hyper‐spectral images [27], enhance the output of semantic segmentation algorithms [13], depth super‐resolution/up‐sampling [11, 13], image colourisation [3, 11, 13], image colour quantisation [12], scale‐space filtering [12, 22], style transfer [10, 12], optical flow estimation [10], compression artefacts removal [14, 22], content‐aware resizing, and stereo matching [10].…”

SRDGAN: learning the noise prior for Super Resolution with Dual Generative Adversarial Networks

“…Various works have been done to recover the details and generate realistic highresolution (HR) images. Recently, learning based methods have shown great potential in solving low-level problems [40,41,15,14]. Especially, development of deep learn-ing provides various learning-based solutions for super resolution [7,8,24,33].…”