Search

In this paper we investigate how distortion and image content interfere within image quality assessment. To this end we analyze how full reference metrics behave within three different groups of images. Given a dataset of images, these are first classified as high, medium or low complexity and the FR methods are applied within each group separately. We consider images from LIVE, CSIQ and LIVE multi-distorted databases. We evaluate 17 full reference quality metrics available in the literature on each of these the high, medium and low complexity groups. We observe that within these groups the metrics better correlate subjective data. In particular, the signal based metrics are the ones that show the highest improvements. Moreover for the LIVE multi-distorted database the gain in performance is evident for all the metrics considered.

In this paper we introduce a multidistortion database, where 10 pristine color images have been simultaneously distorted by two types of distortions: blur and JPEG and noise and JPEG. The two datasets consist of respectively 350 and 400 images, and have been subjectively evaluated within two psycho-physical experiments. We here also propose two no reference multidistortion metrics, one for each of the two datasets, as linear combinations of no reference single distortion ones. The optimized weights of the combinations are obtained using particle swarm optimization. The different combinations proposed show good performance when correlated with the subjective scores of the multidistortion database.

The aim of this work is to study image complexity perception of real images. We conducted psycho-physical experiments where observers judged the complexity of different datasets of images on a web-based interface [1]. At the end of the test, observers indicated the main characteristics that guided their judgements. The databases differed in the type of visual stimuli used: images representing real scenes and/or texture patches. For real scenes the most relevant criteria used were quantity of objects, details and colors, while for texture patches they were regularity and understandability. Several criteria are adopted simultaneously, confirming the multidimensional aspect of complexity found in the literature [2]. To process the subjective data we applied z-scores and outlier removal. The mean scores are then correlated with different visual features. We considered features based on spatial, color and frequency properties that can be associated to bottom-up processes. To take into account top-down effects like understandability we included a memorability index [3]. We propose an image complexity measure where the features are linearly combined. The optimal weighting coefficients are those that best fit the subjective data and depend on the type of stimuli considered. Our measure, properly tuned, can predict complexity perception of different kind of images, outperforming the single visual features. From our investigation two aspects of image complexity can be underlined: many different perceptual properties are involved and their relative influence depends on the type of stimuli. These considerations are supported by both our computational proposal and the verbal description analysis. [1] Ciocca G, Corchs S, Gasparini F, Bricolo E, Tebano R. Does color influence image complexity perception? In: Fifth IAPR Computational Color Imaging Workshop vol. 9016 of Lecture Notes in Computer Science. Springer Berlin/Heidelberg; ((2015) ):139–148 [2] Oliva A, Mack ML, Shrestha M.

Several studies exist in the literature that address the problem of emotion classification of visual stimuli but less effort has been devoted to emotion classification of audio stimuli. The most of these studies start from the analysis of physiological signals such as EEG data [1]. The aim of this work is to evaluate if it is possible to classify audio signals according to elicited emotions using only objective features. In our analysis we adopt the IADS (International Affective Digitized Sound) database [2], composed of 167 auditory stimuli. The database provides pleasure, arousal and dominance ratings for each audio stimulus, recorded from 100 subjects during psycho physical test. The database is formed by different type of audio: from environmental sounds to music, as well as from single sound to complex ones. We start considering the affective dimension of valence within the three categorical classes of low, medium and high pleasure. To investigate this classification task we consider 35 features both in time and frequency domain. With these features, we test three types of classifiers: Bayesian, K Nearest Neighbor and Classification and Regression Tree [3]. We apply a feature selection strategy in order to find the more significant features. Using these features and the Bayesian classifier we have reached an average accuracy of 45%. A similar result is achieved using physiological signals [1]. Starting from our results we believe that dividing each audio files in frames and applying a windowing strategy to evaluate objective features, the final classification performance could significantly increase.

We adopt genetic programming (GP) to define a measure that can predict complexity perception of texture images. We perform psychophysical experiments on three different datasets to collect data on the perceived complexity. The subjective data are used for training, validation, and test of the proposed measure. These data are also used to evaluate several possible candidate measures of texture complexity related to both low level and high level image features. We select four of them (namely roughness, number of regions, chroma variance, and memorability) to be combined in a GP framework. This approach allows a nonlinear combination of the measures and could give hints on how the related image features interact in complexity perception. The proposed complexity measure MGP exhibits Pearson correlation coefficients of 0.890 on the training set, 0.728 on the validation set, and 0.724 on the test set. MGP outperforms each of all the single measures considered. From the statistical analysis of different GP candidate solutions, we found that the roughness measure evaluated on the gray level image is the most dominant one, followed by the memorability, the number of regions, and finally the chroma variance.

The aim of this work is to predict the complexity perception of real world images.We propose a new complexity measure where different image features, based on spatial, frequency and color properties are linearly combined. In order to find the optimal set of weighting coefficients we have applied a Particle Swarm Optimization. The optimal linear combination is the one that best fits the subjective data obtained in an experiment where observers evaluate the complexity of real world scenes on a web-based interface. To test the proposed complexity measure we have performed a second experiment on a different database of real world scenes, where the linear combination previously obtained is correlated with the new subjective data. Our complexity measure outperforms not only each single visual feature but also two visual clutter measures frequently used in the literature to predict image complexity. To analyze the usefulness of our proposal, we have also considered two different sets of stimuli composed of real texture images. Tuning the parameters of our measure for this kind of stimuli, we have obtained a linear combination that still outperforms the single measures. In conclusion our measure, properly tuned, can predict complexity perception of different kind of images.

In this paper we investigate if color influences the perception of image complexity. To this end we perform two different types of psycho-physical experiments on color and grayscale images. In the first experiment, images are ranked based on their complexity (image ranking), while in the second experiment the complexity of each image is assessed on a continuous scale (image scaling). Moreover, we investigate if ten image features, that measure colors as well as other spatial properties of the images, correlate with the collected subjective data. The performance of these correlations are evaluated in terms of Pearson correlation coefficients and Spearman rank-order correlation coefficients. We observe that for each type of experiment, subjective scores for color images are highly correlated with those of the corresponding grayscale versions suggesting that color is not a relevant attribute in evaluating image complexity. Moreover none of the tested simple image features seem to be adapt to predict the image complexity according to the human judgments.

Visual complexity perception plays an important role in the fields of both psychology and computer vision: it can be useful not only to investigate human perception but also to better understand the properties of the objects being perceived. In this paper we investigate the complexity perception of texture images. To this end we perform a psycho-physical experiment on real texture patches. The complexity of each image is assessed on a continuous scale. At the end of the evaluation, each observer indicates the criteria used to assess texture complexity. The most frequent criteria used are regularity, understandability, familiarity and edge density. As candidate complexity measures we consider thirteen image features and we correlate each of them with the subjective scores collected during the experiment. The performance of these correlations are evaluated in terms of Pearson correlation coefficients. The four measures that show the highest correlations are energy, edge density, compression ratio and a visual clutter measure, in accordance with the verbal descriptions collected by the questionnaire.

This thesis focuses on No Reference (NR) methods for Image Quality Assessment (IQA). A review of the IQA field is presented in Chapter 2; where the different IQA methods are described and classified. In particular, the application of IQA methods within a workflow chain is discussed. In Chapter 3 we focus on NR metrics for JPEG-blockiness and noise artifacts. It is in general assumed that subjective methods produce an actual estimate of the perceived quality while objective methods produce values that should be correlated with human perceptions as best as possible. From the analysis of the regression curves that correlate objective and subjective data we have found that in some cases the metric's predictions are not in correspondence with the subjective scores. After reviewing the available databases, we realize that the distortion ranges considered are not in general representative of real case applications. Therefore, in Chapter 4 the Imaging and Vision Lab (IVL) database is introduced. It was generated with the aim of assessing the quality of images corrupted by JPEG and noise. In Chapter 5 we approach the NR-IQA field by focusing on a classification problem. A framework based on machine learning classification is proposed that let us evaluate how images can be classified within different groups or classes, according to their quality. NR metrics are considered as features and the assigned classes are obtained from the psychovisual data. For the JPEG distortion case, the feature space of the classifiers is built using each NR metric as single feature and also a pool of eleven NR metrics. Classification within five and three classes was addressed. In the former case, the five classes are in correspondence to the five categories recommended by the ITU (excellent, good, fair, poor, and bad) when designing image quality experiments. In the latter case we were interested in classifying images as high, medium or low quality ones. The classifiers are trained and tested on

The aim of this work is to study image quality of both single and multiply distorted images. We address the case of images corrupted by Gaussian noise or JPEG compressed as single distortion cases and images corrupted by Gaussian noise and then JPEG compressed, as multiply distortion case. Subjective studies were conducted in two parts to obtain human judgments on the single and multiply distorted images. We study how these subjective data correlate with No Reference state-of-the-art quality metrics. We also investigate proper combining of No Reference metrics to achieve better performance. Results are analyzed and compared in terms of correlation coefficients.

In this paper, we address the Image Quality Assessment (IQA) of JPEG-distorted images. We approach the IQA field by focusing on a classification problem that maps different objective metrics into different categorical quality classes. To this end, we adopt a machine learning classification approach, where No Reference (NR) metrics are considered as features, while the assigned classes come from psycho-visual experiments. Eleven NR metrics have been considered: seven specific for blockiness and four general purpose. We evaluate the performance of single metrics and investigate if a pool of metrics can reach better performances than each of the single ones. Five as well as three quality classes are considered, and the corresponding classifiers are tested on two well known databases available in the literature (LIVE and MICT), and on a new database (IVL) presented in this paper.

Image quality assessment (IQA) is a multi-dimensional research problem and an active and evolving research area. This paper aims to provide an overview of the state of the art of the IQA methods, putting in evidence their applicability and limitations in different application domains. We outline the relationship between the image workflow chain and the IQA approaches reviewing the literature on IQA methods, classifying and summarizing the available metrics. We present general guidelines for three workflow chains in which IQA policies are required. The three workflow chains refer to: high-quality image archives, biometric system and consumer collections of personal photos. Finally, we illustrate a real case study referring to a printing workflow chain, where we suggest and actually evaluate the performance of a set of specific IQA methods.

The aim of our research is to specify experimentally and further model spatial frequency response functions, which quantify human sensitivity to spatial information in real complex images. Three visual response functions are measured: the isolated Contrast Sensitivity Function (iCSF), which describes the ability of the visual system to detect any spatial signal in a given spatial frequency octave in isolation, the contextual Contrast Sensitivity Function (cCSF), which describes the ability of the v isual system to detect a spatial signal in a given octave in an image and the contextual Visual Perception Function (VPF), which describes visual sensitivity to changes in suprathreshold contrast in an image. In this paper we present relevant background, along with our first attempts to derive experimentally and further model the VPF and CSFs. We examine the contrast detection and discrimination frameworks developed by Barten, which we find prov ide a sound starting position for our own modeling purposes. Progress is presented in the following areas: verification of the chosen model for detection and discrimination; choice of contrast metrics for defining contrast sensitivity; apparatus, laboratory set-up and imaging system characterization; stimuli acquisition and stimuli variations; spatial decomposition; methodology for subjective tests. Initial iCSFs are presented and compared with 'classical' findings that hav e used simple visual stimuli, as well as with more recent relevant work in the literature. © 2013 SPIE-IS&T.

We address the problem of image quality assessment for natural images, focusing on No Reference (NR) assessment methods for sharpness. The metrics proposed in the literature are based on edge pixel measures that significantly suffer the presence of noise. In this work we present an automatic method that selects edge segments, making it possible to evaluate sharpness on more reliable data. To reduce the noise influence, we also propose a new sharpness metric for natural images. © 2012 SPIE-IS&T.

In the present article we focus on enhancing the contrast of images with low illumination that present large underexposed regions. Most of these images represent night images. When applying standard contrast enhancement techniques, usually the night mood is modified, and also a noise over-enhancement within the darker regions is introduced. In a previous work we have described our local contrast correction algorithm designed to enhance images where both underexposed and overexposed regions are simoultaneously present. Here we show how this algorithm is able to automatically enhance night images, preserving the original mood. To further improve the performance of our method we also propose here a denoising procedure where the strength of the smoothing is a function of an estimated level of noise and it is further weighted by a saliency map. The method has been applied to a proper database of outdoor and indoor underexposed images. Our results have been qualitatively compared with well know contrast correction methods. © 2011 Springer-Verlag Berlin Heidelberg.

In this work we propose an image quality assessment tool. The tool is composed of different modules that implement several No Reference (NR) metrics (i.e. where the original or ideal image is not available). Different types of image quality attributes can be taken into account by the NR methods, like blurriness, graininess, blockiness, lack of contrast and lack of saturation or colorfulness among others. Our tool aims to give a structured view of a collection of objective metrics that are available for the different distortions within an integrated framework. As each metric corresponds to a single module, our tool can be easily extended to include new metrics or to substitute some of them. The software permits to apply the metrics not only globally but also locally to different regions of interest of the image. © 2011 SPIE-IS&T.

In the present article we focus on enhancing the contrast of images with low illumination that present large underexposed regions. For these particular images, when applying the standard contrast enhancement techniques, we also introduce noise over-enhancement within the darker regions. Even if both the contrast enhancement and denoising problems have been widely addressed within the literature, these two processing steps are, in general, independently considered in the processing pipeline. The goal of this work is to integrate contrast enhancement and denoise algorithms to proper enhance the above described type of images. The method has been applied to a proper database of underexposed images. Our results have been qualitatively compared before and after applying the proposed algorithm. © 2011 SPIE-IS&T.

A method for contrast enhancement is proposed. The algorithm is based on a local and image-dependent exponential correction. The technique aims to correct images that simultaneously present overexposed and underexposed regions. To prevent halo artifacts, the bilateral filter is used as the mask of the exponential correction. Depending on the characteristics of the image (piloted by histogram analysis), an automated parameter-tuning step is introduced, followed by stretching, clipping, and saturation preserving treatments. Comparisons with other contrast enhancement techniques are presented. The Mean Opinion Score (MOS) experiment on grayscale images gives the greatest preference score for our algorithm

The underwater image processing area has received considerable attention within the last decades, showing important achievements. In this paper we review some of the most recent methods that have been specifically developed for the underwater environment. These techniques are capable of extending the range of underwater imaging, improving image contrast and resolution. After considering the basic physics of the light propagation in the water medium, we focus on the different algorithms available in the literature. The conditions for which each of them have been originally developed are highlighted as well as the quality assessment methods used to evaluate their performance. Copyright © 2010 Sheng-Fu Liang et al.

Skin detection is a preliminary step in many applications. We analyze some of the most frequently cited binary skin classifiers based on explicit color cluster definition and present possible strategies to improve their performance. In particular, we demonstrate how this can be accomplished by using genetic algorithms to redefine the cluster boundaries. We also show that the fitness function can be tuned to favor either recall or precision in pixel classification. Some combining strategies are then proposed to further improve the performance of these binary classifiers in terms of recall or precision. Finally, we show that, whatever the method or the strategy employed, the performance can be enhanced by preprocessing the images with a white balance algorithm. All the experiments reported here have been run on a large and heterogeneous image database

A luminance intensity of pixels of an input digital image is corrected for generating a corrected digital image. A luminance of each pixel is calculated as a function of the luminance of a corresponding pixel in an original image according to a parametric function. A mask of the input digital image to be corrected is generated according to a bilateral filtering technique. For each pixel of the input digital image, a respective value of at least one parameter of the parametric function is established based upon the luminance of a corresponding pixel of the mask.

The original dynamic range of a scene is generally constrained into the smaller dynamic range of the acquisition system. This results in the difficulty to design a global tone correction that permits to discriminate both shadow and highlight details. In this work we present a local contrast correction algorithm that allows for simultaneous shadow and highlight adjustments, starting from a simple pixel-wise "gamma" correction, piloted by image statistics analysis

In this paper we examine the performance, on a large and heterogeneous image database, of various skin detectors based on explicit colour skin cluster definition, coupled with a cast remover to see whether, and to what degree, the effectiveness of classification is improved, regardless of the strategy adopted. We also evaluate the hypothesis that a combination of some of the skin detection algorithms studied could ensure a more accurate classification than any of the algorithms provides individually. Different combination rules have been investigated. All the experiments have been performed on the Compaq skin database. The results are evaluated in terms of both recall (the ratio between the number of skin pixels correctly classified and the total number of actual skin pixels), and precision (the ratio between the number of skin pixels correctly classified and the total number of pixels labelled as skin pixels by the detection method employed).

A new approach for selective edge enhancement using unsharp masking is presented. This is based on the premise that biological vision and image reproduction share common principles. In the traditional approach the high frequency components of the image are emphasized, adding to the signal a constant fraction of its high-pass filtered version. The presence of a linear high-pass filter makes the system extremely sensitive to noise. In our approach, the high frequencies added to input image are weighted by a topographic map corresponding to visually salient regions, obtained by a neurodynamical model of visual attention. In this way, the unsharp masking algorithm becomes local and adaptive, enhancing differently the edges according to human perception.

Skin detection is a preliminary step in several applications, and many different methods are available in the literature. We show that the performance of explicit skin cluster classifiers can be enhanced by preprocessing the images with a white balance algorithm. Different combining strategies are then applied to these binary classifiers to further improve their performance in terms of recall and/or precision. Experimental results on a large and heterogeneous image database are presented.

We propose a new approach to select the representative frames for video summarization. The representative frames are selected based on the results of the analysis of the events depicted in the shot in terms of Regions of Interest (ROIs). These ROIs are obtained from a biologically based computational model of visual attention. To select the video frames, that will be part of the final visual summary, we exploit an adaptive temporal sampling method that analyzes the visual feature distribution of the ROIs. Preliminary results are presented and discussed.