Dong-O Kim

Structural information-based image quality assessment using LU factorization

The goal of the objective image quality assessment is to quantitatively measure the image quality of an arbitrary image. The objective image quality measure is desirable if it is close to the subjective image quality assessment such as the mean opinion score. Image quality assessment algorithms are generally classified into two methodologies: perceptual and structural information-based. This paper proposes a structural information-based image quality assessment algorithm, in which LU factorization is used for representation of the structural information of an image. The proposed algorithm performs LU factorization of each of reference and distorted images, from which the distortion map is computed for measuring the quality of the distorted image. Finally, the proposed image quality metric is computed from the two-dimensional distortion map. Experimental results with the laboratory for image and video engineering database images show the efficiency of the proposed method, calibrated by linear and logistic regressions, in terms of the Pearson correlation coefficient and root mean square error. In commercial systems, the proposed algorithm can be used for quality assessment of mobile contents and video coding, which effectively replaces the peak signal to noise ratio or the mean square error.

New Image Quality Metric Using the Harris Response

Image degradation deforms the structure of an image. Thus, to evaluate the quality of an image is equivalent to measuring the degree of deformation of the structure of the image. In this letter, we propose a new image quality metric based on the Harris response, which is computed from the gradient information matrix and its eigenvalues. When an image is degraded by image compression, noise, transmission error, and so on, gradient information of the image is changed, causing the Harris response to change. Therefore, the degree of change in the Harris response of the image is related to the quality degradation of the image. Experimental results with the laboratory for image and video engineering data set show the effectiveness of the proposed quality metric.

Gradient information-based image quality metric

In this paper, we propose a new image quality metric using the gradient information. When an image is degraded, the difference exists between the reference and distorted images. This difference is an important factor in image quality assessment. To assess the quality of an image, we use gradient information of the pixels having large differences between the reference and distorted images. In this paper, the Harris response (HR), a well-known feature, is used to obtain the gradient information for assessing the image quality. That is, HR values at pixels having the nonzero difference between the reference and distorted images are compared for evaluating the image quality. For detecting these pixels, we use a cross-projection tensor based edge suppression technique. Experimental results with the LIVE data set show the effectiveness of the proposed quality measure.

Detection of moving edges based on the concept of entropy and cross-entropy

We propose a moving edge extraction using the concept of entropy and cross-entropy, in which the cross-entropy concept is applied to dynamic scene analysis. The cross- entropy concept provides enhancement of detection for the dynamically changed area. We combine the results of cross- entropy in the difference picture (DP) with those of entropy in the current frame so that we can effectively extract moving edges. We also propose the moving edge extraction method by combining the results of cross-entropy and those of Laplacian of Gaussian (LoG).

Image quality measure using the phase quantization code

In this paper, we propose a new image quality measure using the phase quantization code (PQC), which represents the sign of phase quadrature output of a complex filter such as Gabor filters. When an image is degraded by image compression, noise, transmission error, and so on, signs of phase quadrature outputs also change. Therefore, degradation of the image quality results in the change of the PQC of the image. Experimental results with the LIVE data set show the effectiveness of the proposed quality measure.

Visual quality metric using one-dimensional histograms of motion vectors

Quality assessment methods are classified into three types depending on the availability of the reference image or video: full-reference (FR), reduced-reference (RR), or no-reference (NR). This paper proposes efficient RR visual quality metrics, called motion vector histogram based quality metrics (MVHQMs). In assessing the visual quality of a video, the overall impression of a video tends to be regarded as the visual quality of the video. To compare two motion vectors (MVs) extracted from reference and distorted videos, we define the one-dimensional (horizontal and vertical) MV histograms as features, which are computed by counting the number of occurrences of MVs over all frames of a video. For testing the similarity between MV histograms, two different MVHQMs using the histogram intersection and histogram difference are proposed. We evaluate the effectiveness of the two proposed MVHQMs by comparing their results with differential mean opinion score (DMOS) data for 46 video clips of common intermediate format (CIF)/quarter CIF (QCIF) that are coded under varying bit rates/frame rates with H.263. We compare the performance of the proposed metrics and conventional quality measures. Experimental results with various test video sequences show that the proposed MVHQMs give better performance than the conventional methods in various aspects such as the performance, stability, and data size.

Image quality assessment using the amplitude/phase quantization code

In our previous work, we presented the phase quantization code-based image quality metric (IQM). Amplitude change as well as phase change is an important factor for evaluating image quality. In this paper, we present a new IQM based on the amplitude/phase quantization code (APQC). When an image is degraded by image compression, noise, transmission error, and so on, the amplitude and phase of quadrature outputs change. In other words, degradation of the image quality results in change in the APQC of the image. Experimental results with the LIVE database show the effectiveness of the proposed quality metric.

New image quality metric using derivative filters and compressive sensing

In this paper, we propose a new image quality metric using derivative filters in the context of compressive sensing (CS) that represents a sparse or compressible signal with a small number of measurements. In general, an arbitrary image is not sparse or compressible, however, its derivative image is compressible. In this paper, derivative images are obtained using first- and second-order derivative filters such as Sobel operators and Laplacian of Gaussian filters. Each derivative image of the reference and distorted images is measured via CS. Measurements of derivative images are compared for evaluating the image quality. Experiments with the laboratory for image and video engineering database show the effectiveness of the proposed method.

Evaluation of the image quality based on random measurements

In this paper, instead of using an arbitrary image feature representing the structure of an image, random measurements are used for evaluating the image quality. Inner products between random bases (consisting of random values) and an image are called random measurements, which impose the information of the image and also change if the image is distorted. Random measurements of the reference and distorted images are compared to evaluate the image quality. Experimental results with the laboratory in image and video engineering image database show the effectiveness of the proposed quality metric using random measurements.

Recognition of 3D objects with curved surfaces based on the cross entropy between shape histograms

In the real world, many objects consist of curved surfaces, thus recognition of 3D objects with curved surfaces is to be investigated. In this paper, we present the shape histogram based algorithm for recognizing and grouping objects, in which the cross entropy between shape histograms is employed. Computer simulations with various synthetic and real images are presented to show the effectiveness of the proposed algorithm.