Image Quality Estimation Research Articles

Application of the Non-Reference Metrics for the Quality Estimation of the Current Images of the Stationary Ground Multi-object Scenes. Part 2

Application of the Non-Reference Metrics for the Quality Estimation of the Current Images of the Stationary Ground Multi-object Scenes. Part 2

PATCH-IQ: A patch based learning framework for blind image quality assessment

Most well-known blind image quality assessment (BIQA) models usually follow a two-stage framework whereby various types of features are first extracted and used as an input to a regressor. The regression algorithm is used to model human perceptual measures based on a training set of distorted images. However, this approach requires an intensive training phase to optimise the regression parameters. In this paper, we overcome this limitation by proposing an alternative BIQA model that predicts image quality using nearest neighbour methods which have virtually zero training cost. The model, termed PATCH based blind Image Quality assessment (PATCH-IQ), has a learning framework that operates at the patch level. This enables PATCH-IQ to provide not only a global image quality estimation but also a local image quality estimation. Based on the assumption that the perceived quality of a distorted image will be best predicted by features drawn from images with the same distortion class, PATCH-IQ also introduces a distortion identification stage in its framework. This enables PATCH-IQ to identify the distortion affecting the image, a property that can be useful for further local processing stages. PATCH-IQ is evaluated on the standard IQA databases, and the provided scores are highly correlated to human perception of image quality. It also delivers competitive prediction accuracy and computational performance in relationship to other state-of-the-art BIQA models.

Application of the Non-reference Metrics for the Quality Estimation of the Current Images of the Multi-object Stationary Ground Scenes. Part 1

Application of the Non-reference Metrics for the Quality Estimation of the Current Images of the Multi-object Stationary Ground Scenes. Part 1

Fully automatic multi-atlas segmentation of CTA for partial volume correction in cardiac SPECT/CT

Anatomical-based partial volume correction (PVC) has been shown to improve image quality and quantitative accuracy in cardiac SPECT/CT. However, this method requires manual segmentation of various organs from contrast-enhanced computed tomography angiography (CTA) data. In order to achieve fully automatic CTA segmentation for clinical translation, we investigated the most common multi-atlas segmentation methods. We also modified the multi-atlas segmentation method by introducing a novel label fusion algorithm for multiple organ segmentation to eliminate overlap and gap voxels. To evaluate our proposed automatic segmentation, eight canine 99mTc-labeled red blood cell SPECT/CT datasets that incorporated PVC were analyzed, using the leave-one-out approach. The Dice similarity coefficient of each organ was computed. Compared to the conventional label fusion method, our proposed label fusion method effectively eliminated gaps and overlaps and improved the CTA segmentation accuracy. The anatomical-based PVC of cardiac SPECT images with automatic multi-atlas segmentation provided consistent image quality and quantitative estimation of intramyocardial blood volume, as compared to those derived using manual segmentation. In conclusion, our proposed automatic multi-atlas segmentation method of CTAs is feasible, practical, and facilitates anatomical-based PVC of cardiac SPECT/CT images.

Investigation of undersampling and reconstruction algorithm dependence on respiratory correlated 4D-MRI for online MR-guided radiation therapy

The purpose of this work is to investigate the effects of undersampling and reconstruction algorithm on the total processing time and image quality of respiratory phase-resolved 4D MRI data. Specifically, the goal is to obtain quality 4D-MRI data with a combined acquisition and reconstruction time of five minutes or less, which we reasoned would be satisfactory for pre-treatment 4D-MRI in online MRI-gRT. A 3D stack-of-stars, self-navigated, 4D-MRI acquisition was used to scan three healthy volunteers at three image resolutions and two scan durations. The NUFFT, CG-SENSE, SPIRiT, and XD-GRASP reconstruction algorithms were used to reconstruct each dataset on a high performance reconstruction computer. The overall image quality, reconstruction time, artifact prevalence, and motion estimates were compared. The CG-SENSE and XD-GRASP reconstructions provided superior image quality over the other algorithms. The combination of a 3D SoS sequence and parallelized reconstruction algorithms using computing hardware more advanced than those typically seen on product MRI scanners, can result in acquisition and reconstruction of high quality respiratory correlated 4D-MRI images in less than five minutes.

Analysis of Geometric and Spatial Image Quality of KOMPSAT-3A Imagery in Comparison with KOMPSAT-3 Imagery

This study investigates the geometric and spatial image quality analysis of KOMPSAT-3A stereo pair. KOMPSAT-3A is, the latest satellite of KOMPSAT family, a Korean earth observation satellite operating in optical bands. A KOMPSAT-3A stereo pair was taken on 23 November, 2015 with 0.55 m ground sampling distance over Terrassa area of Spain. The convergence angle of KOMPSAT-3A stereo pair was estimated as 58.68˚. The investigation was assessed through the evaluation of the geopositioning analysis, image quality estimation and the accuracy of automatic Digital Surface Model (DSM) generation and compared with those of KOMPSAT-3 stereo pair with the convergence angle of 44.80˚ over the same area. First, geopositioning accuracy was tested with initial rational polynomial coefficients (RPCs) and after compensating the biases of the initial RPCs by manually collected ground control points. Then, regarding image quality, relative edge response was estimated for manually selected points visible from two stereo pairs. Both of the initial and biascompensated positioning accuracy and the quality assessment result expressed that KOMPSAT-3A images showed higher performance than those of KOMPSAT-3 images. Finally, the accuracy of DSMs generated from KOMPSAT-3A and KOMPSAT-3 stereo pairs were examined with respect to the reference LiDAR-derived DSM. The various DSMs were generated over the whole coverage of individual stereo pairs with different grid spacing and over three types of terrain; flat, mountainous and urban area. Root mean square errors of DSM from KOMPSAT-3A pair were larger than those for KOMPSAT-3. This seems due to larger convergence angle of the KOMPSAT-3A stereo pair.

MS-UNIQUE: Multi-model and Sharpness-weighted Unsupervised Image Quality Estimation

In this paper, we train independent linear decoder models to estimate the perceived quality of images. More specifically, we calculate the responses of individual non-overlapping image patches to each of the decoders and scale these responses based on the sharpness characteristics of filter set. We use multiple linear decoders to capture different abstraction levels of the image patches. Training each model is carried out on 100,000 image patches from the ImageNet database in an unsupervised fashion. Color space selection and ZCA Whitening are performed over these patches to enhance the descriptiveness of the data. The proposed quality estimator is tested on the LIVE and the TID 2013 image quality assessment databases. Performance of the proposed method is compared against eleven other state of the art methods in terms of accuracy, consistency, linearity, and monotonic behavior. Based on experimental results, the proposed method is generally among the top performing quality estimators in all categories.

Estimation of Fingerprint Image Quality in Accordance with Photographing Conditions

Estimation of Fingerprint Image Quality in Accordance with Photographing Conditions

Wide-aperture aspherical lens for high-resolution terahertz imaging.

In this paper, we introduce wide-aperture aspherical lens for high-resolution terahertz (THz) imaging. The lens has been designed and analyzed by numerical methods of geometrical optics and electrodynamics. It has been made of high-density polyethylene by shaping at computer-controlled lathe and characterized using a continuous-wave THz imaging setup based on a backward-wave oscillator and Golay detector. The concept of image contrast has been implemented to estimate image quality. According to the experimental data, the lens allows resolving two points spaced at 0.95λ distance with a contrast of 15%. To highlight high resolution in the THz images, the wide-aperture lens has been employed for studying printed electronic circuit board containing sub-wavelength-scale elements. The observed results justify the high efficiency of the proposed lens design.

Improving Performance of IQA Algorithm: A Simplified Approach to SSIM based on Image Gradients

Objectives: To propose a fast and efficient algorithm for Full Reference Image Quality Assessment (FR-IQA) that can evaluate the quality of a distorted image accurately. Methods/Statistical Analysis: The proposed approach uses SSIM (Structural Similarity Index ) algorithm based on image gradients. SSIM algorithm has been widely accepted as an effective tool for estimating image quality. The quality estimation capability of SSIM has been enhanced by using image gradients. In this paper, the SSIM based on image gradients has been simplified and optimized to reduce the computational complexity, thereby to improve the execution speed and to improve the quality estimation capability. Findings: The proposed algorithm is computationally simple compared to SSIM and Gradient based SSIM. There has been significant improvement in the execution speed and quality prediction capability of the proposed algorithm compared to SSIM and Gradient based SSIM for various types of distortions. This simplified version of SSIM based on gradients has been extensively tested on popular image databases. The results confirm its effectiveness, efficiency and consistency in estimating the image quality. Applications/Improvements: The proposed algorithm is suitable in IQA applications where improved prediction accuracy and execution speed are important.

Full-Reference Video Quality Estimation for Videos With Different Spatial Resolutions

Full-reference (FR) video quality estimators (QEs) resize either the distorted input video or the reference video to compute the quality when these videos have different spatial resolutions. This resizing operation causes several limitations. Multiscale Image Quality Estimator (MIQE) overcomes those limitations for images, but it does not consider the temporal characteristics of video. In this paper, we develop an FR video QE that integrates MIQE with the motion information to estimate the quality of the distorted video without resampling the reference or the test videos. We also perform subjective tests to compare the proposed algorithm with the existing QEs. In these tests, the reference and the input videos are displayed at their native resolutions. The test results show that the proposed algorithm outperforms other QEs when the reference video and the input video have different spatial resolutions. We have also evaluated the performance of the approach using the Scalable Video Database.

UNIQUE: Unsupervised Image Quality Estimation

In this paper, we estimate perceived image quality using sparse representations obtained from generic image databases through an unsupervised learning approach. A color space transformation, a mean subtraction, and a whitening operation are used to enhance descriptiveness of images by reducing spatial redundancy; a linear decoder is used to obtain sparse representations; and a thresholding stage is used to formulate suppression mechanisms in a visual system. A linear decoder is trained with 7 GB worth of data, which corresponds to 100,000 8x8 image patches randomly obtained from nearly 1,000 images in the ImageNet 2013 database. A patch-wise training approach is preferred to maintain local information. The proposed quality estimator UNIQUE is tested on the LIVE, the Multiply Distorted LIVE, and the TID 2013 databases and compared with thirteen quality estimators. Experimental results show that UNIQUE is generally a top performing quality estimator in terms of accuracy, consistency, linearity, and monotonic behavior.

CSV: Image quality assessment based on color, structure, and visual system

This paper presents a full-reference image quality estimator based on color, structure, and visual system characteristics denoted as CSV. In contrast to the majority of existing methods, we quantify perceptual color degradations rather than absolute pixel-wise changes. We use the CIEDE2000, color difference formulation to quantify low-level color degradations and the Earth Mover's Distance between color name probability vectors to measure significant color degradations. In addition to the perceptual color difference, CSV also contains structural and perceptual differences. Structural feature maps are obtained by mean subtraction and divisive normalization, and perceptual feature maps are obtained from contrast sensitivity formulations of retinal ganglion cells. The proposed quality estimator CSV is tested on the LIVE, the Multiply Distorted LIVE, and the TID 2013 databases, and it is always among the top two performing quality estimators in terms of at least ranking, monotonic behavior or linearity.

Imaging quality assessment of mammograms

The aim. The aim of the work is imaging quality assessment of pathological structures on the low-contrast gray-scale images of the mammary gland by developing a new method for multicriteria estimation of the mammograms quality.Solving methods. A method for multicriteria estimation of the visualization mammograms quality was developed. It is based on expert assessments and taking into account both subjective and objective characteristics of the mammographic images.Different criteria for optimality in the conditions of uncertainty is proposed to use for the comparison of average expert estimates of the mammographic image quality resulting from the different processing methods.Results. The proposed method for multicriteria evaluation of the visualization mammograms quality has been tested on real digital mammography images, which were obtained with the help of mammography complex Syma.The results of a comparative study of mammograms before and using various digital processing techniques have shown that the most optimal for radiologist perception and correct diagnosis is mammography image, which is processed by a method for enhancing imaging quality mammograms IMRI-MAM.Conclusions Thus, the application of the method for multicriteria evaluation of the visualization mammograms quality with the help of integrated scales allows to quickly and accurately carry out a visual assessment of the mammographic image quality. The proposed method will get rid of the time-consuming and inaccurate assessment of the modern imaging techniques only by subjective assessment.

Reduced-Reference Quality Assessment Based on the Entropy of DWT Coefficients of Locally Weighted Gradient Magnitudes.

Perceptual image quality assessment (IQA) attempts to use computational models to estimate the image quality in accordance with subjective evaluations. Reduced-reference (RR) image quality assessment (IQA) methods make use of partial information or features extracted from the reference image for estimating the quality of distorted images. Finding a balance between the number of RR features and accuracy of the estimated image quality is essential and important in IQA. In this paper we propose a training-free low-cost RRIQA method that requires a very small number of RR features (6 RR features). The proposed RRIQA algorithm is based on the discrete wavelet transform (DWT) of locally weighted gradient magnitudes.We apply human visual system's contrast sensitivity and neighborhood gradient information to weight the gradient magnitudes in a locally adaptive manner. The RR features are computed by measuring the entropy of each DWT subband, for each scale, and pooling the subband entropies along all orientations, resulting in L RR features (one average entropy per scale) for an L-level DWT. Extensive experiments performed on seven large-scale benchmark databases demonstrate that the proposed RRIQA method delivers highly competitive performance as compared to the state-of-the-art RRIQA models as well as full reference ones for both natural and texture images. The MATLAB source code of REDLOG and the evaluation results are publicly available online at https://http://lab.engineering.asu.edu/ivulab/software/redlog/.

Individual Differences in Image-Quality Estimations

Subjective image-quality estimation with high-quality images is often a preference-estimation task. Preferences are subjective, and individual differences exist. Individual differences are also seen in the eye movements of people. A task's subjectivity can result from people using different rules as a basis for their estimation. Using two studies, we investigated whether different preference-estimation rules are related to individual differences in viewing behaviour by examining the process of preference estimation of high-quality images. The estimation rules were measured from free subjective reports on important quality-related attributes (Study 1) and from estimations of the attributes’ importance in preference estimation (Study 2). The free reports showed that the observers used both feature-based image-quality attributes (e.g., sharpness, illumination) and abstract attributes, which include an interpretation of the image features (e.g., atmosphere and naturalness). In addition, the observers were classified into three viewing-strategy groups differing in fixation durations in both studies. These groups also used different estimation rules. In both studies, the group with medium-length fixations differed in their estimation rules from the other groups. In Study 1, the observers in this group used more abstract attributes than those in the other groups; in Study 2, they considered atmosphere to be a more important image feature. The study shows that individual differences in a quality-estimation task are related to both estimation rules and viewing strategies, and that the difference is related to the level of abstraction of the estimations.

Parameter-based estimation of CT dose index and image quality using an in-house android™-based software

Compromise between radiation dose and image quality is essential in the use of CT imaging. CT dose index (CTDI) is currently the primary dosimetric formalisms in CT scan, while the low and high contrast resolutions are aspects indicating the image quality. This study was aimed to estimate CTDIvol and image quality measures through a range of exposure parameters variation. CTDI measurements were performed using PMMA (polymethyl methacrylate) phantom of 16 cm diameter, while the image quality test was conducted by using catphan ® 600. CTDI measurements were carried out according to IAEA TRS 457 protocol using axial scan mode, under varied parameters of tube voltage, collimation or slice thickness, and tube current. Image quality test was conducted accordingly under the same exposure parameters with CTDI measurements. An Android™ based software was also result of this study. The software was designed to estimate the value of CTDIvol with maximum difference compared to actual CTDIvol measurement of 8.97%. Image quality can also be estimated through CNR parameter with maximum difference to actual CNR measurement of 21.65%.

64 - Dose and image quality estimation in computed tomography

64 - Dose and image quality estimation in computed tomography

Phase retrieval from a single fringe pattern by using empirical wavelet transform

Phase retrieval from a single fringe pattern is one of the key tasks in optical metrology. In this paper, we present a new method for phase retrieval from a single fringe pattern based on empirical wavelet transform. In the proposed method, a fringe pattern can be effectively divided into three components: nonuniform background, fringes and random noise, which are described in different sub-pass. So the phase distribution information can be robustly extracted from fringes representing a fundamental frequency component. In simulation and a practical projection fringes test, the performance of the present method is successfully verified by comparing with the conventional wavelet transform method in terms of both image quality and phase estimation errors.

Digital Image Quality Assessment Based on Standard Normal Deviation

We propose a new method that specifies objective image quality factors by evaluating an image quality measurement model using random images. In other words, No-Reference variables are used to evaluate the quality of an original image without using any reference for comparison. 1000 portrait images were collected from a web gallery with votes constituting over 30 recommendation values. The bottom-up data collecting process was used to calculate the following image quality factors: total range, average, standard deviation, normalized distribution, z-score, preference percentage. A final grade is awarded out of 100 points, and this method ranks and grades the final estimated image quality preference in terms of total image quality factors. The results of the proposed image quality evaluation model consist of the specific dynamic range, skin tone R, G, B, L, A, B, and RSC contrast. We can present the total for the expected preference points as the average of the objective image qualities. Our proposed image quality evaluation model can measure the preferences for an actual image using a statistical analysis. The results indicate that this is a practical image quality measurement model that can extract a subjects preferred image quality.