Image quality assessment using deep convolutional networks

Abstract

Image quality assessment metric for frame accumulated image Review of Scientific Instruments 89, 013703 (2018); https://doi.org/10.1063/1.5020715Image super resolution using deep convolutional network based on topology aggregation structure AIP Conference Proceedings 1864, 020185 (2017); https://doi.org/10.1063/1.4993002 AIP ADVANCES 7, 125324 (2017) Image quality assessment using deep convolutional networks This paper proposes a method of accurately assessing image quality without a reference image by using a deep convolutional neural network. Existing training based methods usually utilize a compact set of linear filters for learning features of images captured by different sensors to assess their quality. These methods may not be able to learn the semantic features that are intimately related with the features used in human subject assessment. Observing this drawback, this work proposes training a deep convolutional neural network (CNN) with labelled images for image quality assessment. The ReLU in the CNN allows non-linear transformations for extracting high-level image features, providing a more reliable assessment of image quality than linear filters. To enable the neural network to take images of any arbitrary size as input, the spatial pyramid pooling (SPP) is introduced connecting the top convolutional layer and the fully-connected layer. In addition, the SPP makes the CNN robust to object deformations to a certain extent. The proposed method taking an image as input carries out an end-to-end learning process, and outputs the quality of the image. It is tested on public datasets. Experimental results show that it outperforms existing methods by a large margin and can accurately assess the image quality on images taken by different sensors of varying sizes. © 2017 Author (s)

Blind image quality assessment based on Benford's law

“…When testing on images in the TID2008 database, we need to convert the TID2008 MOS to the same scale as the DMOS. Similar treatment of the MOS is reported in [61]. Table 15 shows the SROCC of our algorithm compared with four state-of-the-art IQA algorithms.…”

An FPGA Implementation of a Convolutional Auto-Encoder

“…According to Equation (1), the image compression ratio for our CAE is calculated. PSNR [26] is often used to assess the quality of signal reconstruction. The formula of PSNR is as follows:…”

Optimal Combination of Image Denoisers

“…Our proposed solution is a deep neural network MSE estimator. Using deep neural networks for image quality assessment is an active research topic [46]- [50]. However, the existing neural network based image quality assessment methods are tailored to predict the human visual system responses when presenting an image to a user.…”

“…The denoising strength is set as one of the values σ = 10, 20, 30, 40, and 50. When testing, we use a noise level of σ ∈ [10,50]. In this experiment, the noise is unclipped i.i.d.…”

“…Figure 2 illustrates the behavior of DnCNN and BM3D as the denoising strength σ deviates from the actual level σ. In this experiment, we use five denoising strengths σ = {10, 20, 30, 40, 50} and a continuous range of σ ∈ [10,50]. As shown in the plot, BM3D has a slightly more robust performance, in the sense that a chosen denoising strength σ can work for a reasonable wide range of actual noise levels σ.…”