• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.1
• /
• pp.288-307
• /
• 2016

Image distortions are typically characterized by degradations of structures. Dictionaries learned from natural images can capture the underlying structures in images, which are important for image quality assessment (IQA). This paper presents a general-purpose no-reference image quality metric using a GRadient-Induced Dictionary (GRID). A dictionary is first constructed based on gradients of natural images using K-means clustering. Then image features are extracted using the dictionary based on Euclidean-norm coding and max-pooling. A distortion classification model and several distortion-specific quality regression models are trained using the support vector machine (SVM) by combining image features with distortion types and subjective scores, respectively. To evaluate the quality of a test image, the distortion classification model is used to determine the probabilities that the image belongs to different kinds of distortions, while the regression models are used to predict the corresponding distortion-specific quality scores. Finally, an overall quality score is computed as the probability-weighted distortion-specific quality scores. The proposed metric can evaluate image quality accurately and efficiently using a small dictionary. The performance of the proposed method is verified on public image quality databases. Experimental results demonstrate that the proposed metric can generate quality scores highly consistent with human perception, and it outperforms the state-of-the-arts.

• Journal of Korea Multimedia Society
• /
• v.7 no.12
• /
• pp.1657-1664
• /
• 2004

We propose a method that classifies images into two object types man-made and natural objects. A central object is extracted from each image by using central object extraction method[1] before classification. A central object in an images defined as a set of regions that lies around center of the image and has significant color distribution against its surrounding. We define three measures to classify the object images. The first measure is energy of edge direction histogram. The energy is calculated based on the direction of only non-circular edges. The second measure is an energy difference along directions in Gabor filter dictionary. Maximum and minimum energy along directions in Gabor filter dictionary are selected and the energy difference is computed as the ratio of the maximum to the minimum value. The last one is a shape of an object, which is also represented by Gabor filter dictionary. Gabor filter dictionary for the shape of an object differs from the one for the texture in an object in which the former is computed from a binarized object image. Each measure is combined by using majority rule tin which decisions are made by the majority. A test with 600 images shows a classification accuracy of 86%.

• IEIE Transactions on Smart Processing and Computing
• /
• v.6 no.4
• /
• pp.262-268
• /
• 2017

Intelligent video surveillance systems have been developed to monitor global areas and find specific target objects using a large-scale database. However, person re-identification presents some challenges, such as pose change and occlusions. To solve the problems, this paper presents an improved person re-identification method using sparse representation and saliency-based dictionary construction. The proposed method consists of three parts: i) feature description based on salient colors and textures for dictionary elements, ii) orthogonal atom selection using cosine similarity to deal with pose and viewpoint change, and iii) measurement of reconstruction error to rank the gallery corresponding a probe object. The proposed method provides good performance, since robust descriptors used as a dictionary atom are generated by weighting some salient features, and dictionary atoms are selected by reducing excessive redundancy causing low accuracy. Therefore, the proposed method can be applied in a large scale-database surveillance system to search for a specific object.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.3
• /
• pp.1700-1721
• /
• 2017

The problem of blocking artifacts is very common in block-based image and video compression, especially at very low bit rates. In this paper, we propose a post-processing method for JPEG-coded image deblocking via sparse representation and adaptive residual threshold. This method includes three steps. First, we obtain the dictionary by online dictionary learning and the compressed images. The dictionary is then modified by the histogram of oriented gradient (HOG) feature descriptor and K-means cluster. Second, an adaptive residual threshold for orthogonal matching pursuit (OMP) is proposed and used for sparse coding by combining blind image blocking assessment. At last, to take advantage of human visual system (HVS), the edge regions of the obtained deblocked image can be further modified by the edge regions of the compressed image. The experimental results show that our proposed method can keep the image more texture and edge information while reducing the image blocking artifacts.

• Journal of Broadcast Engineering
• /
• v.20 no.5
• /
• pp.676-686
• /
• 2015

A 4D light field image is represented in traditional 2D spatial domain and additional 2D angular domain. The 4D light field has a resolution limitation both in spatial and angular domains since 4D signals are captured by 2D CMOS sensor with limited resolution. In this paper, we propose a dictionary learning-based superresolution algorithm in 4D light field domain to overcome the resolution limitation. The proposed algorithm performs dictionary learning using a large number of extracted 4D light field patches. Then, a high resolution light field image is reconstructed from a low resolution input using the learned dictionary. In this paper, we reconstruct a 4D light field image to have double resolution both in spatial and angular domains. Experimental result shows that the proposed method outperforms the traditional method for the test images captured by a commercial light field camera, i.e. Lytro.

• ETRI Journal
• /
• v.32 no.4
• /
• pp.596-602
• /
• 2010

This paper proposes a computationally efficient learning-based super-resolution algorithm using k-means clustering. Conventional learning-based super-resolution requires a huge dictionary for reliable performance, which brings about a tremendous memory cost as well as a burdensome matching computation. In order to overcome this problem, the proposed algorithm significantly reduces the size of the trained dictionary by properly clustering similar patches at the learning phase. Experimental results show that the proposed algorithm provides superior visual quality to the conventional algorithms, while needing much less computational complexity.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.5
• /
• pp.2539-2554
• /
• 2017

Regression-based image super resolution (SR) methods have shown great advantage in time consumption while maintaining similar or improved quality performance compared to other learning-based methods. In this paper, we propose a novel single image SR method based on hierarchical regression to further improve the quality performance. As an improvement to other regression-based methods, we introduce a hierarchical scheme into the process of learning multiple regressors. First, training samples are grouped into different clusters according to their geometry similarity, which generates the structure layer. Then in each cluster, a compact dictionary can be learned by Sparse Coding (SC) method and the training samples can be further grouped by dictionary atoms to form the detail layer. Last, a series of projection matrixes, which anchored to dictionary atoms, can be learned by linear regression. Experiment results show that hierarchical scheme can lead to regression that is more precise. Our method achieves superior high quality results compared with several state-of-the-art methods.

• Journal of Korea Multimedia Society
• /
• v.9 no.9
• /
• pp.1208-1221
• /
• 2006

In this paper, we implement an AAC(Augmentative and Alternative Communication) system based on image dictionary in order that speech defectives can easily communicate their opinion to others by using images. Normally, those who have a speech defect use only a few limited words to express their intentions and it is an effective way to use images in their communication because they have a difficulty in speaking. Therefore we make verbal images of verb and adjective that play an important role in expressing the speaker's intention, define pattern of semantic relation between the verbal images and noun images, and construct the image dictionary. In this AAC system, when a user clicks a verbal image, the system generates a sentence by selecting noun images which are component parts of corresponding pattern based on semantic relation with the verbal image. We evaluate the implemented system by how efficiently children of speech defect can express their intention and the result shows more than 70% of success rate in communication.

• Journal of Information Processing Systems
• /
• v.17 no.2
• /
• pp.411-425
• /
• 2021

In this paper, a vehicle recognition algorithm based on deep convolutional neural network and compression dictionary is proposed. Firstly, the network structure of fine vehicle recognition based on convolutional neural network is introduced. Then, a vehicle recognition system based on multi-scale pyramid convolutional neural network is constructed. The contribution of different networks to the recognition results is adjusted by the adaptive fusion method that adjusts the network according to the recognition accuracy of a single network. The proportion of output in the network output of the entire multiscale network. Then, the compressed dictionary learning and the data dimension reduction are carried out using the effective block structure method combined with very sparse random projection matrix, which solves the computational complexity caused by high-dimensional features and shortens the dictionary learning time. Finally, the sparse representation classification method is used to realize vehicle type recognition. The experimental results show that the detection effect of the proposed algorithm is stable in sunny, cloudy and rainy weather, and it has strong adaptability to typical application scenarios such as occlusion and blurring, with an average recognition rate of more than 95%.

• Journal of the Korea Convergence Society
• /
• v.11 no.3
• /
• pp.19-26
• /
• 2020

In this paper, we propose a method to analyze the loss region of the dictionary-based super resolution result learned for image quality improvement and to map the learning data according to the analyzed loss region. In the conventional learned dictionary-based method, a result different from the feature configuration of the input image may be generated according to the learning image, and an unintended artifact may occur. The proposed method estimate loss information of low resolution images by analyzing the reconstructed contents to reduce inconsistent feature composition and unintended artifacts in the example-based super resolution process. By mapping the training data according to the final interpolation feature map, which improves the noise and pixel imbalance of the estimated loss information using a Gaussian-based kernel, it generates super resolution with improved noise, artifacts, and staircase compared to the existing super resolution. For the evaluation, the results of the existing super resolution generation algorithms and the proposed method are compared with the high-definition image, which is 4% better in the PSNR (Peak Signal to Noise Ratio) and 3% in the SSIM (Structural SIMilarity Index).