• The Journal of the Korea Contents Association
• /
• v.6 no.9
• /
• pp.19-27
• /
• 2006

Mean shift procedure is applied for the data points in the joint spatial-range domain and achieves a high quality. However, a color image is segmented differently according to the inputted spatial parameter or range parameter and the demerit is that the image is broken into many small regions in case of the small parameter. In this paper, to improve this demerit, we propose the method that groups similar regions using region merging method for over-segmented images. The proposed method converts a over-segmented image in RGB color space into in HSI color space and merges similar regions by hue information. Here, to preserve edge information, the region merge constraints are used to decide whether regions are merged or not. After then, we merge the regions in RGB color space for non-processed regions in HSI color space. Experimental results show the superiority in region's segmentation results.

• Proceedings of the KAIS Fall Conference
• /
• 2003.11a
• /
• pp.223-230
• /
• 2003

Feature-based similarity retrieval has become an important research issue in multimedia database systems. The features of multimedia data are useful for discriminating between multimedia objects (e 'g', documents, images, video, music score, etc.). For example, images are represented by their color histograms, texture vectors, and shape descriptors, and are usually high-dimensional data. The performance of conventional multidimensional data structures(e'g', R- Tree family, K-D-B tree, grid file, TV-tree) tends to deteriorate as the number of dimensions of feature vectors increases. The R＊-tree is the most successful variant of the R-tree. In this paper, we propose a SOM-based R＊-tree as a new indexing method for high-dimensional feature vectors.The SOM-based R＊-tree combines SOM and R＊-tree to achieve search performance more scalable to high dimensionalities. Self-Organizing Maps (SOMs) provide mapping from high-dimensional feature vectors onto a two dimensional space. The mapping preserves the topology of the feature vectors. The map is called a topological of the feature map, and preserves the mutual relationship (similarity) in the feature spaces of input data, clustering mutually similar feature vectors in neighboring nodes. Each node of the topological feature map holds a codebook vector. A best-matching-image-list. (BMIL) holds similar images that are closest to each codebook vector. In a topological feature map, there are empty nodes in which no image is classified. When we build an R＊-tree, we use codebook vectors of topological feature map which eliminates the empty nodes that cause unnecessary disk access and degrade retrieval performance. We experimentally compare the retrieval time cost of a SOM-based R＊-tree with that of an SOM and an R＊-tree using color feature vectors extracted from 40, 000 images. The result show that the SOM-based R＊-tree outperforms both the SOM and R＊-tree due to the reduction of the number of nodes required to build R＊-tree and retrieval time cost.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.4
• /
• pp.273-280
• /
• 2013

This paper presents novel OCS-LBP (Oriented Center Symmetric Local Binary Patterns) based on orientation of pixel gradient and image retrieval system based on BoF (Bag-of-Feature) and random forest classifier. Feature vectors extracted from training data are clustered into code book and each feature is transformed new BoF feature using code book. BoF features are applied to random forest for training and random forest having N classes is constructed by combining several decision trees. For testing, the same OCS-LBP feature is extracted from a query image and BoF is applied to trained random forest classifier. In contrast to conventional retrieval system, query image selects similar K-nearest neighbor (K-NN) classes after random forest is performed. Then, Top K similar images are retrieved from database images that are only labeled K-NN classes. Compared with other retrieval algorithms, the proposed method shows both fast processing time and improved retrieval performance.

• Restorative Dentistry and Endodontics
• /
• v.38 no.3
• /
• pp.146-153
• /
• 2013

Objectives: Aluminum step wedge (ASW) equivalent radiodensity (eRD) has been used to quantify restorative material's radiodensity. The aim of this study was to evaluate the effects of image acquisition control (IAC) of a digital X-ray system on the radiodensity quantification under different exposure time settings. Materials and Methods: Three 1-mm thick restorative material samples with various opacities were prepared. Samples were radiographed alongside an ASW using one of three digital radiographic modes (linear mapping (L), nonlinear mapping (N), and nonlinear mapping and automatic exposure control activated (E)) under 3 exposure time settings (underexposure, normal-exposure, and overexposure). The ASW eRD of restorative materials, attenuation coefficients and contrasts of ASW, and the correlation coefficient of linear relationship between logarithms of gray-scale value and thicknesses of ASW were compared under 9 conditions. Results: The ASW eRD measurements of restorative materials by three digital radiographic modes were statistically different (p = 0.049) but clinically similar. The relationship between logarithms of background corrected grey scale value and thickness of ASW was highly linear but attenuation coefficients and contrasts varied significantly among 3 radiographic modes. Varying exposure times did not affect ASW eRD significantly. Conclusions: Even though different digital radiographic modes induced large variation on attenuation of coefficient and contrast of ASW, E mode improved diagnostic quality of the image significantly under the underexposure condition by improving contrasts, while maintaining ASW eRDs of restorative materials similar. Under the condition of this study, underexposure time may be acceptable clinically with digital X-ray system using automatic gain control that reduces radiation exposure for patient.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.10
• /
• pp.3708-3728
• /
• 2021

Indian herbal plants are used in agriculture and in the food, cosmetics, and pharmaceutical industries. Laboratory-based tests are routinely used to identify and classify similar herb species by analyzing their internal cell structures. In this paper, we have applied computer vision techniques to do the same. The original leaf image was preprocessed using the Chan-Vese active contour segmentation algorithm to efface the background from the image by setting the contraction bias as (v) -1 and smoothing factor (µ) as 0.5, and bringing the initial contour close to the image boundary. Thereafter the segmented grayscale image was fed to a leaky capacitance fired neuron model (LCFN), which differentiates between similar herbs by combining different groups of pixels in the leaf image. The LFCN's decay constant (f), decay constant (g) and threshold (h) parameters were empirically assigned as 0.7, 0.6 and h=18 to generate the 1D feature vector. The LCFN time sequence identified the internal leaf structure at different iterations. Our proposed framework was tested against newly collected herbal species of natural images, geometrically variant images in terms of size, orientation and position. The 1D sequence and shape features of aloe, betel, Indian borage, bittergourd, grape, insulin herb, guava, mango, nilavembu, nithiyakalyani, sweet basil and pomegranate were fed into the 5-fold Bayesian regularization neural network (BRNN), K-nearest neighbors (KNN), support vector machine (SVM), and ensemble classifier to obtain the highest classification accuracy of 91.19%.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.10
• /
• pp.1317-1322
• /
• 2021

Nonlocal similarity of natural images leads to the fact that a patch matrix whose columns are similar patches of the reference patch has a low rank. Images corrupted by additive white Gaussian noises (AWGN) make their patch matrices to have a higher rank. The noise in the image can be reduced by obtaining low rank approximation of the patch matrices. In this paper, an image denoising algorithm is proposed, which first constructs the patch matrices by combining the similar patches of each reference patch, which is a part of the noisy image. For each patch matrix, the proposed algorithm calculates its low rank approximate, and then recovers the original image by aggregating the low rank estimates. The simulation results using widely accepted test images show that the proposed denoising algorithm outperforms four recent methods.

• Journal of Cardiovascular Imaging
• /
• v.31 no.2
• /
• pp.108-115
• /
• 2023

BACKGROUND: Minimizing contrast dose and radiation exposure while maintaining image quality during computed tomography angiography (CTA) for transcatheter aortic valve replacement (TAVR) is desirable, but not well established. This systematic review compares image quality for low contrast and low kV CTA versus conventional CTA in patients with aortic stenosis undergoing TAVR planning. METHODS: We performed a systematic literature review to identify clinical studies comparing imaging strategies for patients with aortic stenosis undergoing TAVR planning. The primary outcomes of image quality as assessed by the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were reported as random effects mean difference with 95% confidence interval (CI). RESULTS: We included 6 studies reporting on 353 patients. There was no difference in cardiac SNR (mean difference, -1.42; 95% CI, -5.71 to 2.88; p = 0.52), cardiac CNR (mean difference, -3.83; 95% CI, -9.98 to 2.32; p = 0.22), aortic SNR (mean difference, -0.23; 95% CI, -7.83 to 7.37; p = 0.95), aortic CNR (mean difference, -3.95; 95% CI, -12.03 to 4.13; p = 0.34), and ileofemoral SNR (mean difference, -6.09; 95% CI, -13.80 to 1.62; p = 0.12) between the low dose and conventional protocols. There was a difference in ileofemoral CNR between the low dose and conventional protocols with a mean difference of -9.26 (95% CI, -15.06 to -3.46; p = 0.002). Overall, subjective image quality was similar between the 2 protocols. CONCLUSIONS: This systematic review suggests that low contrast and low kV CTA for TAVR planning provides similar image quality to conventional CTA.

• International Journal of Highway Engineering
• /
• v.18 no.6
• /
• pp.145-152
• /
• 2016

OBJECTIVES : Visibility at night can be improved by using retroreflection for short distances and phosphorescent line markings for long distances. In this study, we analyzed the characteristics of the phosphorescent line marking through a laboratory luminance test. Field performance analysis was performed through tests conducted on the road. We also examined the luminance measurement methods using the digital image obtained during the phosphorescent visibility evaluation. METHODS : In this study, the laboratory luminance test of the phosphorescent line marking was conducted using seven specimens to characterize the luminance changes according to the type of the glass beads, the thickness of the phosphorescent line marking, and the brightness and irradiation time of the light source. Phosphorescent and general line markings were made at 150 m to investigate the field luminance performance. A preliminary review of the luminance measurement methods was made using a digital image from a digital single-lens reflex (DSLR) camera. The measured luminance ratio of the general and the phosphorescent line markings was compared with the calculated luminance ratio using luminance analysis. RESULTS : Through the laboratory luminance test, it was seen that the change in luminance, which corresponds to the brightness of the light source, appears large but the influence of the thickness and irradiation time is low. The field performance test of the phosphorescent line marking conducted on the road involved measuring the luminance on the day the marking was made and 7 days after the marking was made. The luminance was found to be $190mcd/m^2$ at 30 min after sunset and approximately $10-12mcd/m^2$ 4h after sunset. The results of the luminance test were captured using a digital image for each time group. The luminance ratio of the phosphorescent line marking, when compared to that of the general line marking, showed a similar trend within a 13% maximum error. Additionally, when this luminance ratio is compared to the direct field measurement, it could be confirmed that the luminance ratio, as captured in the digital image, showed a similar tendency. CONCLUSIONS : 1) The change in luminance corresponding to the brightness of the light source is significant in comparison with that corresponding to the thickness and the irradiation time. In addition, the results of the field test for the phosphorescent line marking satisfied the phosphorescent fire protection standard. 2) We examined the validity of the luminance measurement method using a digital image and we concluded that the change in the luminance ratio shows a similar tendency in both the cases. The results can form the basis for luminance measurement methodology for the construction and maintenance of phosphorescent line markings.

• Journal of Internet Computing and Services
• /
• v.22 no.1
• /
• pp.23-30
• /
• 2021

This paper proposes a noise-tolerant image classification system using multiple autoencoders. The development of deep learning technology has dramatically improved the performance of image classifiers. However, if the images are contaminated by noise, the performance degrades rapidly. Noise added to the image is inevitably generated in the process of obtaining and transmitting the image. Therefore, in order to use the classifier in a real environment, we have to deal with the noise. On the other hand, the autoencoder is an artificial neural network model that is trained to have similar input and output values. If the input data is similar to the training data, the error between the input data and output data of the autoencoder will be small. However, if the input data is not similar to the training data, the error will be large. The proposed system uses the relationship between the input data and the output data of the autoencoder, and it has two phases to classify the images. In the first phase, the classes with the highest likelihood of classification are selected and subject to the procedure again in the second phase. For the performance analysis of the proposed system, classification accuracy was tested on a Gaussian noise-contaminated MNIST dataset. As a result of the experiment, it was confirmed that the proposed system in the noisy environment has higher accuracy than the CNN-based classification technique.

• Science of Emotion and Sensibility
• /
• v.12 no.1
• /
• pp.129-136
• /
• 2009

Unlike former researches, this study for developing the caricature generator began observing the methods that other caricature experts have adopted. According to the observation, it seemed that experts tried to exaggerate characteristics of the target shape from other similar objects. When we are saying "This is similar to that," we give salience to their difference among the identical form groups. This study was to find the most similar geometry form to the target shape and then to transform its form through exaggeration. The research scope was restricted to exaggerate the outline shape of two-dimensional looped curve as a caricature form. For this, the author discussed the following: (a) organization method of four kinds of similar geometry form database, (b) search method to find the pertinent similar geometry form, (c) arrangement method for those searched data, and (d) method to exaggerate the target shape. Human faces and cars were selected as research categories to make the database. According to the survey over the transformed results, it was proved its possibility.