Channel (digital image)

About: Channel (digital image) is a research topic. Over the lifetime, 7211 publications have been published within this topic receiving 69974 citations.

Depth estimation from a monocular view of the outdoors

Abstract: This study presents an algorithm for estimating three-dimensional (3D) depth from a single monocular outdoor image. This algorithm combines three major components: the initial depth prior, the dark channel prior, and multi-resolution analysis. The initial depth prior assigns an initial depth to each object based on its location in the image. The object segmentation in this step is based on edge, hue, and saturation. The dark channel prior cooperates with color saturation to generate approximate depth information. Combining the depth maps of both initial depth prior and dark channel prior generates depth maps of three resolutions. Fusing these maps together across different resolutions produces the final depth map. Experimental results show that the proposed method achieves a considerable computational savings compared to existing methods while retaining a depth map close to the ground truth.

Pre-Attention and Spatial Dependency Driven No-Reference Image Quality Assessment

Abstract: The excessive emulation of the human visual system and the lack of connection between chromatic data and distortion have been the major bottlenecks in developing image quality assessment. To address this issue, we develop a new no-reference (NR) image quality assessment (IQA) metric that accounts for the impact of pre-attention and spatial dependency on the perceived quality of distorted images. The resulting model, dubbed the Pre-attention and Spatial-dependency driven Quality Assessment (PSQA) predictor, introduces the pre-attention theory to emulate early phase visual perception by refining luminance-channel data. Chromatic data are also processed concurrently by transforming images from RGB to the perceptually optimized SCIELAB color space. Considering that the gray-tone spatial dependency matrix conveys important texture properties that are closely related to visual quality, this matrix, as a mathematical solution for subsequent visual process emulation, is calculated along with its statistical features on both gray and color channels. To clarify the influence of different regression procedures on model output, support vector regression and AdaBoosting Back Propagation (BP) neural networks are adopted separately to train the prediction models. We thoroughly evaluated PSQA on four public image quality databases: LIVE, TID2013, CSIQ, and VCL. The experimental results show that PSQA delivers highly competitive performance compared with top-rank NR and full-reference IQA metrics.

Fog removal in images using improved dark channel prior and contrast limited adaptive histogram equalization

Abstract: It is necessary to perform fog removal from an image based on the estimation of depth to increase the visibility of a scene. In this paper, we propose a new algorithm to eradicate fog from images in which fog is defined as a state or cause of perplexity or confusion with respect to the image. It runs at high speed and simultaneously minimizes the halo-artifact with a new median operator in dark channel prior. The proposed method is based on Guided Filter for transmission-map refinement and Contrast Limited Adaptive Histogram Equalization (CLAHE) for visibility improvement. It preserves small details while remaining robust against density of fog, and recovers scene contrast simultaneously. Guided filter improved the transmission map acquired from Median dark channel prior (MDCP), which is an improvement of the Dark Channel Prior DCP by the use of median operation. All of the parameters used in our method are data driven. The quality of algorithm has been validated on several types of fog-degraded images where considerable variation in contrast and illumination exists. Moreover, its performance is compared with the other state-of-the-art methods. The experimental results indicate that the proposed method effectively restores the color and contrast of scene as well as produces satisfactory information in homogeneous fog. It outperforms the existing fog removal methods for run time computational time and other evaluation metrics for rating of visibility enhancement. The proposed method conserves small details part of the image when outstanding vigorous against concentration of fog, and recuperate scene contrast instantaneously. It controls at a high speed than the existing approaches and can diminish the halo effect.