Christophe Charrier

Blind Image Quality Assessment: A Natural Scene Statistics Approach in the DCT Domain

We develop an efficient general-purpose blind/no-reference image quality assessment (IQA) algorithm using a natural scene statistics (NSS) model of discrete cosine transform (DCT) coefficients. The algorithm is computationally appealing, given the availability of platforms optimized for DCT computation. The approach relies on a simple Bayesian inference model to predict image quality scores given certain extracted features. The features are based on an NSS model of the image DCT coefficients. The estimated parameters of the model are utilized to form features that are indicative of perceptual quality. These features are used in a simple Bayesian inference approach to predict quality scores. The resulting algorithm, which we name BLIINDS-II, requires minimal training and adopts a simple probabilistic model for score prediction. Given the extracted features from a test image, the quality score that maximizes the probability of the empirically determined inference model is chosen as the predicted quality score of that image. When tested on the LIVE IQA database, BLIINDS-II is shown to correlate highly with human judgments of quality, at a level that is competitive with the popular SSIM index.

A DCT Statistics-Based Blind Image Quality Index

The development of general-purpose no-reference approaches to image quality assessment still lags recent advances in full-reference methods. Additionally, most no-reference or blind approaches are distortion-specific, meaning they assess only a specific type of distortion assumed present in the test image (such as blockiness, blur, or ringing). This limits their application domain. Other approaches rely on training a machine learning algorithm. These methods however, are only as effective as the features used to train their learning machines. Towards ameliorating this we introduce the BLIINDS index (BLind Image Integrity Notator using DCT Statistics) which is a no-reference approach to image quality assessment that does not assume a specific type of distortion of the image. It is based on predicting image quality based on observing the statistics of local discrete cosine transform coefficients, and it requires only minimal training. The method is shown to correlate highly with human perception of quality.

Blind Prediction of Natural Video Quality

We propose a blind (no reference or NR) video quality evaluation model that is nondistortion specific. The approach relies on a spatio-temporal model of video scenes in the discrete cosine transform domain, and on a model that characterizes the type of motion occurring in the scenes, to predict video quality. We use the models to define video statistics and perceptual features that are the basis of a video quality assessment (VQA) algorithm that does not require the presence of a pristine video to compare against in order to predict a perceptual quality score. The contributions of this paper are threefold. 1) We propose a spatio-temporal natural scene statistics (NSS) model for videos. 2) We propose a motion model that quantifies motion coherency in video scenes. 3) We show that the proposed NSS and motion coherency models are appropriate for quality assessment of videos, and we utilize them to design a blind VQA algorithm that correlates highly with human judgments of quality. The proposed algorithm, called video BLIINDS, is tested on the LIVE VQA database and on the EPFL-PoliMi video database and shown to perform close to the level of top performing reduced and full reference VQA algorithms.

DCT statistics model-based blind image quality assessment

We propose an efficient, general-purpose, distortion-agnostic, blind/no-reference image quality assessment (NR-IQA) algorithm based on a natural scene statistics model of discrete cosine transform (DCT) coefficients. The algorithm is computationally appealing, given the availability of platforms optimized for DCT computation. We propose a generalized parametric model of the extracted DCT coefficients. The parameters of the model are utilized to predict image quality scores. The resulting algorithm, which we name BLIINDS-II, requires minimal training and adopts a simple probabilistic model for score prediction. When tested on the LIVE IQA database, BLIINDS-II is shown to correlate highly with human visual perception of quality, at a level that is even competitive with the powerful full-reference SSIM index.

Maximum likelihood difference scaling of image quality in compression-degraded images

Lossy image compression techniques allow arbitrarily high compression rates but at the price of poor image quality. We applied maximum likelihood difference scaling to evaluate image quality of nine images, each compressed via vector quantization to ten different levels, within two different color spaces, RGB and CIE 1976 L*a*b*. In L*a*b* space, images could be compressed on average by 32% more than in RGB space, with little additional loss in quality. Further compression led to marked perceptual changes. Our approach permits a rapid, direct measurement of the consequences of image compression for human observers.

Color image segmentation using morphological clustering and fusion with automatic scale selection

In this paper, a color image segmentation method considering pairwise color projections is proposed. Each pairwise projection is analyzed according to an unsupervised morphological clustering which looks for the dominant colors of a 2D histogram. This leads to obtaining three segmentation maps combined by superposition after being simplified. The superposition process itself producing an over-segmentation of the image, a pairwise region merging is performed according to a similarity criterion up to a termination criterion. To fully automate the segmentation, an energy function is proposed to quantify the segmentation quality. The latter acts as a performance indicator and is used all over the segmentation to tune its parameters: the scale of the unsupervised morphological clustering and the termination criterion of region merging. Experimental results are conducted on a reference image database and comparisons with state-of-the-art algorithms.

Machine learning to design full-reference image quality assessment algorithm

A crucial step in image compression is the evaluation of its performance, and more precisely, available ways to measure the quality of compressed images. In this paper, a machine learning expert, providing a quality score is proposed. This quality measure is based on a learned classification process in order to respect human observers. The proposed method namely Machine Learning-based Image Quality Measure (MLIQM) first classifies the quality using multi-Support Vector Machine (SVM) classification according to the quality scale recommended by the ITU. This quality scale contains 5 ranks ordered from 1 (the worst quality) to 5 (the best quality). To evaluate the quality of images, a feature vector containing visual attributes describing images content is constructed. Then, a classification process is performed to provide the final quality class of the considered image. Finally, once a quality class is associated to the considered image, a specific SVM regression is performed to score its quality. Obtained results are compared to the one obtained applying classical Full-Reference Image Quality Assessment (FR-IQA) algorithms to judge the efficiency of the proposed method.

SVM training time reduction using vector quantization

In this paper, we describe a new method for training SVM on large data sets. Vector quantization is applied to reduce a large data set by replacing examples by prototypes. Training time for choosing optimal parameters is greatly reduced. Some experimental results yields to demonstrate that this method can reduce training time by a factor of 100, while preserving classification rate. Moreover this method allows to find a decision function with a low complexity when the training data set includes noisy or error examples.

Combination of Multiple Pixel Classifiers for Microscopic Image Segmentation

The combination of classifiers has been proposed as a method allowing to improve the quality of recognition sy stems as compared to a single classifier. This paper describes a segmentation scheme based on a combination of pixel classifications. The aim of this paper is to show the influence of the neighborhood information and of the number of classifier used in several combination processes. In the first part, we detail the ground of our study for a microscopic application . Then, we name the different steps of the new segmentation scheme. In the third and fourth part, we detail the different rules that can be used to combine classifiers and the classific tions results obtained on colour microscopic images. Final ly, we draw a conclusion on the improvement of the quality of the segmentation at the end of treatment.

Fast pixel classification by SVM using vector quantization, tabu search and hybrid color space

In this paper, a new learning method is proposed to build Support Vector Machines (SVM) Binary Decision Function (BDF) of reduced complexity, efficient generalization and using an adapted hybrid color space. The aim is to build a fast and efficient SVM classifier of pixels. The Vector Quantization (VQ) is used in our learning method to simplify the training set. This simplification step maps pixels of the training set to representative prototypes. A criterion is defined to evaluate the Decision Function Quality (DFQ) which blends recognition rate and complexity of a BDF. A model selection based on the selection of the simplification level, of a hybrid color space and of SVM hyperparameters is performed to optimize this DFQ. Search space for selecting the best model being huge. Our learning method uses Tabu Search (TS) metaheuritics to find a good sub-optimal model on tractable times. Experimental results show the efficiency of the method.