scispace - formally typeset
Search or ask a question
Author

Hsin-Hua Liu

Bio: Hsin-Hua Liu is an academic researcher from National Taiwan University. The author has contributed to research in topics: Image quality & Feature (computer vision). The author has an hindex of 11, co-authored 20 publications receiving 216 citations.

Papers
More filters
Proceedings ArticleDOI
01 Sep 2019
TL;DR: Dense blocks for the U-Net architecture are introduced, which can alleviate the problem of gradient disappearance, while also reducing the number of parameters.
Abstract: Deep convolutional neural networks (DCNN) have demonstrated their potential to generate reasonable results in image inpainting. Some existing method uses convolution to generate surrounding features, then passes features by fully connected layers, and finally predicts missing regions. Although the final result is semantically reasonable, some blurred situations generated because the standard convolution is used, which conditioned on the effective pixels and the substitute values in the masked holes. In this paper, we introduce dense blocks for the U-Net architecture, which can alleviate the problem of gradient disappearance, while also reducing the number of parameters. The most important is that it can enhance the transfer of features and make more efficient use of them. Partial convolution is used to solve the problem of artifacts such as color differences and blurring. Experiments on the place365 dataset demonstrate our approach can generate more detailed and semantically reasonable results in random area image inpainting.

33 citations

Journal ArticleDOI
TL;DR: This paper creates a high-definition image database, which has higher resolution than most of the image quality databases, and collects 250 source images from 10 categories, which are far more diversified than other existing quality databases.
Abstract: In this paper, we focus on the creation of general purpose 2-D image quality databases. Although there are many of them, they still lack some important characteristics, such as high-definition resolution, diversified source images, more commonly seen distortions, and a larger amount of test (distorted) images. To tackle this problem, we create a high-definition image database, which has higher resolution than most of the image quality databases. In addition, we collect 250 source images from 10 categories, which are far more diversified than other existing quality databases. Moreover, we generate 10 most commonly seen distortions to represent the real world scenario. Finally, 12000 test images are generated for the whole database, which is the largest data set so far compared with other general purpose image quality databases with human subjective ratings. The subjective test is conducted in a controlled environment to obtain the ground truth of image quality, where we collect over 360000 opinion scores. We believe the birth of this quality database would help further development of research on image quality assessment.

27 citations

Journal ArticleDOI
TL;DR: The experimental results show that the proposed model leads to significant performance improvement on quality prediction for stereoscopic images compared with other existing state-of-the-art quality metrics.
Abstract: We proposed a blind image quality assessment model which used classification and prediction for three-dimensional (3D) image quality assessment (denoted as CAP-3DIQA) that can automatically evaluate the quality of stereoscopic images. First, in the classification stage, the model separated the distorted images into several subsets according to the types of image distortions. This process will assign the images with the same distortion type to the same group. After the classification stage, the classified distorted image set is fed into the image quality predictor that contains five different perceptual channels which predict the image quality score individually. Finally, we used the regression module of the support vector machine to evaluate the final image quality score, where the input of the regression model is the combination of five channel’s outputs. The model, we proposed is tested on three public and popular databases, which are LIVE 3D Image Quality Database Phase I, LIVE 3D Image Quality Database Phase II, and MCL 3D Image Quality Database. The experimental results show that our proposed model leads to significant performance improvement on quality prediction for stereoscopic images compared with other existing state-of-the-art quality metrics.

27 citations

Proceedings ArticleDOI
19 Jul 2010
TL;DR: After some experiments, this proposed quality assessment metric does work well in the Laboratory for Image and Video Engineering (LIVE) Video Quality Database and is also competitive with the other existing state-of-the-art video quality assessment methods.
Abstract: This paper proposed a new objective video quality metric for multimedia videos based on a different perspective. We extended one existing image quality assessment metric to a video quality metric by considering temporal information and converted it into some compensation factor to correct the video quality score obtained in the spatial domain. After some experiments, we find out this proposed quality assessment metric does work well in the Laboratory for Image and Video Engineering (LIVE) Video Quality Database and is also competitive with the other existing state-of-the-art video quality assessment methods.

27 citations

Proceedings ArticleDOI
19 Aug 2016
TL;DR: A new divide-and-conquer based method, called fusion of multiple binary age-grouping-estimation systems, for human facial age estimation, which can achieve satisfying results and outperform other state-of-the-art age estimation approaches.
Abstract: In this paper, we propose a new divide-and-conquer based method, called fusion of multiple binary age-grouping-estimation systems, for human facial age estimation. Under a specific constraint, such as a given facial feature or classification/regression method, what is the better framework for age estimation? First we employ multiple binary-grouping systems for age group classification. Each face image will be classified into one of the two groups. Within the two groups, two models are trained to estimate ages for the faces classified into their groups, respectively. We also investigate the effect of different age grouping systems on the performance of age grouping accuracy and age estimation error. In the last stage, we propose a sequentially selection algorithm to fuse some of the binary-grouping systems to get a final age estimation result. Experiments on the MORPH2 database demonstrate our framework for age estimation can achieve satisfying results and outperform other state-of-the-art age estimation approaches.

27 citations


Cited by
More filters
Christopher M. Bishop1
01 Jan 2006
TL;DR: Probability distributions of linear models for regression and classification are given in this article, along with a discussion of combining models and combining models in the context of machine learning and classification.
Abstract: Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

10,141 citations

Book ChapterDOI
E.R. Davies1
01 Jan 1990
TL;DR: This chapter introduces the subject of statistical pattern recognition (SPR) by considering how features are defined and emphasizes that the nearest neighbor algorithm achieves error rates comparable with those of an ideal Bayes’ classifier.
Abstract: This chapter introduces the subject of statistical pattern recognition (SPR). It starts by considering how features are defined and emphasizes that the nearest neighbor algorithm achieves error rates comparable with those of an ideal Bayes’ classifier. The concepts of an optimal number of features, representativeness of the training data, and the need to avoid overfitting to the training data are stressed. The chapter shows that methods such as the support vector machine and artificial neural networks are subject to these same training limitations, although each has its advantages. For neural networks, the multilayer perceptron architecture and back-propagation algorithm are described. The chapter distinguishes between supervised and unsupervised learning, demonstrating the advantages of the latter and showing how methods such as clustering and principal components analysis fit into the SPR framework. The chapter also defines the receiver operating characteristic, which allows an optimum balance between false positives and false negatives to be achieved.

1,189 citations

Journal ArticleDOI
14 Aug 1987-JAMA
TL;DR: Although a variety of univariate statistics are included, certain topics that are important in medical research are not, and there is little or no discussion of multiple regression, life-table techniques, or pooling of studies.
Abstract: This book attempts to achieve a difficult goal: to teach statistics to the novice so as to impart a liking and understanding of statistics. The book is geared toward a medical audience, since most examples are from the medical literature. The structure of the book consists of the following elements in each chapter: a small number of statistical rules of thumb, followed by a nontechnical explanation, a demonstration of how to work through the mechanics of doing the statistical test in question, a summary, and sample problems to be solved by the reader. (The answers, with explanations, are provided in an appendix.) Although a variety of univariate statistics are included, certain topics that are important in medical research are not. For example, there is little or no discussion of multiple regression, life-table techniques, or pooling of studies. These omissions, especially of multiple regression, are unfortunate. The Primer was derived from

898 citations

Journal ArticleDOI
TL;DR: Experimental results show that the proposed blind image watermarking scheme has stronger robustness against most common attacks such as image compression, filtering, cropping, noise adding, blurring, scaling and sharpening etc.

157 citations