Compact Ultra-Wideband Wilkinson Power Divider in Parallel Stripline with Modified Isolation Branches.

An efficient design method for a compact and ultra-wideband multi-stage Wilkinson power divider in a parallel stripline (PSL) is proposed. To enhance the frequency bandwidth of the proposed power divider while reducing its size, the isolation branch is modified; that is, two capacitors are connected to both sides of a resistor at each isolation branch. For an efficient design process, the PSL power divider is equivalently represented by two microstrip power dividers, and the design equations are derived. Based on the design equations, an in-house algorithm is utilized to optimally determine the design parameters, including the line impedance, resistance, and capacitance of each stage. For example, a three-stage PSL power divider is designed with three λ/4 transmission lines at a base frequency of 5 GHz. To verify the accuracy of the design procedure, 3D EM simulations and measurements are performed, and the results show good agreement. Compared with the conventional three-stage Wilkinson power divider, the proposed PSL power divider achieves a wider frequency bandwidth of 1.16 to 6.51 GHz (139.5%) and a 23% shorter transmission line length of 207°, while exhibiting an insertion loss of 0.7 to 1.4 dB.

Ultra-Wideband Vertical Transition in Coplanar Stripline for Ultra-High-Speed Digital Interfaces.

A design method for an ultra-wideband coplanar-stripline-based vertical transition that can be used for ultra-high-speed digital interfaces is proposed. A conventional via structure, based on a differential line (DL), inherently possesses performance limitations (<10 GHz) due to difficulties in maintaining constant line impedance and smooth electric field transformation, in addition to the effects of signal skews, FR4 fiber weave, and unbalanced EM interferences. DL-based digital interfaces may not meet the demands of ultra-high-speed digital data transmission required for the upcoming 6G communications. The use of a coplanar stripline (CPS), a type of planar balanced line (BL), for the vertical transition, along with the ultra-wideband DL-to-CPS transition, mostly removes the inherent and unfavorable issues of the DL and enables ultra-high-speed digital data transmission. The design process of the transition is simplified using the analytical design formulas, derived using the conformal mapping method, of the transition. The characteristic line impedances of the transition are calculated and found to be in close agreement with the results obtained from EM simulations. Utilizing these results, the CPS-based vertical transition, maintaining the characteristic line impedance of 100 Ω, is designed and fabricated. The measured results confirm its ultra-wideband characteristics, with a maximum of 1.6 dB insertion loss and more than 10 dB return loss in the frequency range of DC to 30 GHz. Therefore, the proposed CPS-based vertical transition offers a significantly wider frequency bandwidth, i.e., more than three times that of conventional DL-based via structures.

Toward Exploiting Second-Order Feature Statistics for Arbitrary Image Style Transfer.

Generating images of artistic style from input images, also known as image style transfer, has been improved in the quality of output style and the speed of image generation since deep neural networks have been applied in the field of computer vision research. However, the previous approaches used feature alignment techniques that were too simple in their transform layer to cover the characteristics of style features of images. In addition, they used an inconsistent combination of transform layers and loss functions in the training phase to embed arbitrary styles in a decoder network. To overcome these shortcomings, the second-order statistics of the encoded features are exploited to build an optimal arbitrary image style transfer technique. First, a new correlation-aware loss and a correlation-aware feature alignment technique are proposed. Using this consistent combination of loss and feature alignment methods strongly matches the second-order statistics of content features to those of the target-style features and, accordingly, the style capacity of the decoder network is increased. Secondly, a new component-wise style controlling method is proposed. This method can generate various styles from one or several style images by using style-specific components from second-order feature statistics. We experimentally prove that the proposed method achieves improvements in both the style capacity of the decoder network and the style variety without losing the ability of real-time processing (less than 200 ms) on Graphics Processing Unit (GPU) devices.

Total Style Transfer with a Single Feed-Forward Network.

The development of recent image style transfer methods allows the quick transformation of an input content image into an arbitrary style. However, these methods have a limitation that the scale-across style pattern of a style image cannot be fully transferred into a content image. In this paper, we propose a new style transfer method, named total style transfer, that resolves this limitation by utilizing intra/inter-scale statistics of multi-scaled feature maps without losing the merits of the existing methods. First, we use a more general feature transform layer that employs intra/inter-scale statistics of multi-scaled feature maps and transforms the multi-scaled style of a content image into that of a style image. Secondly, we generate a multi-scaled stylized image by using only a single decoder network with skip-connections, in which multi-scaled features are merged. Finally, we optimize the style loss for the decoder network in the intra/inter-scale statistics of image style. Our improved total style transfer can generate a stylized image with a scale-across style pattern from a pair of content and style images in one forwarding pass. Our method achieved less memory consumption and faster feed-forwarding speed compared with the recent cascade scheme and the lowest style loss among the recent style transfer methods.

Compact Image-Style Transfer: Channel Pruning on the Single Training of a Network.

Recent image-style transfer methods use the structure of a VGG feature network to encode and decode the feature map of the image. Since the network is designed for the general image-classification task, it has a number of channels and, accordingly, requires a huge amount of memory and high computational power, which is not mandatory for such a relatively simple task as image-style transfer. In this paper, we propose a new technique to size down the previously used style transfer network for eliminating the redundancy of the VGG feature network in memory consumption and computational cost. Our method automatically finds a number of consistently inactive convolution channels during the network training phase by using two new losses, i.e., channel loss and xor loss. The former maximizes the number of inactive channels and the latter fixes the positions of these inactive channels to be the same for the image. Our method improves the image generation speed to be up to 49% faster and reduces the number of parameters by 20% while maintaining style transferring performance. Additionally, our losses are also effective in pruning the VGG16 classifier network, i.e., parameter reduction by 26% and top-1 accuracy improvement by 0.16% on CIFAR-10.

Arbitrary Font Generation by Encoder Learning of Disentangled Features.

Making a new font requires graphical designs for all base characters, and this designing process consumes lots of time and human resources. Especially for languages including a large number of combinations of consonants and vowels, it is a heavy burden to design all such combinations independently. Automatic font generation methods have been proposed to reduce this labor-intensive design problem. Most of the methods are GAN-based approaches, and they are limited to generate the trained fonts. In some previous methods, they used two encoders, one for content, the other for style, but their disentanglement of content and style is not sufficiently effective in generating arbitrary fonts. Arbitrary font generation is a challenging task because learning text and font design separately from given font images is very difficult, where the font images have both text content and font style in each image. In this paper, we propose a new automatic font generation method to solve this disentanglement problem. First, we use two stacked inputs, i.e., images with the same text but different font style as content input and images with the same font style but different text as style input. Second, we propose new consistency losses that force any combination of encoded features of the stacked inputs to have the same values. In our experiments, we proved that our method can extract consistent features of text contents and font styles by separating content and style encoders and this works well for generating unseen font design from a small number of reference font images that are human-designed. Comparing to the previous methods, the font designs generated with our method showed better quality both qualitatively and quantitatively than those with the previous methods for Korean, Chinese, and English characters. e.g., 17.84 lower FID in unseen font compared to other methods.

Part Affinity Fields and CoordConv for Detecting Landmarks of Lumbar Vertebrae and Sacrum in X-ray Images.

With the prevalence of degenerative diseases due to the increase in the aging population, we have encountered many spine-related disorders. Since the spine is a crucial part of the body, fast and accurate diagnosis is critically important. Generally, clinicians use X-ray images to diagnose the spine, but X-ray images are commonly occluded by the shadows of some bones, making it hard to identify the whole spine. Therefore, recently, various deep-learning-based spinal X-ray image analysis approaches have been proposed to help diagnose the spine. However, these approaches did not consider the characteristics of frequent occlusion in the X-ray image and the properties of the vertebra shape. Therefore, based on the X-ray image properties and vertebra shape, we present a novel landmark detection network specialized in lumbar X-ray images. The proposed network consists of two stages: The first step detects the centers of the lumbar vertebrae and the upper end plate of the first sacral vertebra (S1), and the second step detects the four corner points of each lumbar vertebra and two corner points of S1 from the image obtained in the first step. We used random spine cutout augmentation in the first step to robustify the network against the commonly obscured X-ray images. Furthermore, in the second step, we used CoordConv to make the network recognize the location distribution of landmarks and part affinity fields to understand the morphological features of the vertebrae, resulting in more accurate landmark detection. The proposed network was evaluated using 304 X-ray images, and it achieved 98.02% accuracy in center detection and 8.34% relative distance error in corner detection. This indicates that our network can detect spinal landmarks reliably enough to support radiologists in analyzing the lumbar X-ray images.

A Review of Multi-Modal Learning from the Text-Guided Visual Processing Viewpoint.

For decades, co-relating different data domains to attain the maximum potential of machines has driven research, especially in neural networks. Similarly, text and visual data (images and videos) are two distinct data domains with extensive research in the past. Recently, using natural language to process 2D or 3D images and videos with the immense power of neural nets has witnessed a promising future. Despite the diverse range of remarkable work in this field, notably in the past few years, rapid improvements have also solved future challenges for researchers. Moreover, the connection between these two domains is mainly subjected to GAN, thus limiting the horizons of this field. This review analyzes Text-to-Image (T2I) synthesis as a broader picture, Text-guided Visual-output (T2Vo), with the primary goal being to highlight the gaps by proposing a more comprehensive taxonomy. We broadly categorize text-guided visual output into three main divisions and meaningful subdivisions by critically examining an extensive body of literature from top-tier computer vision venues and closely related fields, such as machine learning and human-computer interaction, aiming at state-of-the-art models with a comparative analysis. This study successively follows previous surveys on T2I, adding value by analogously evaluating the diverse range of existing methods, including different generative models, several types of visual output, critical examination of various approaches, and highlighting the shortcomings, suggesting the future direction of research.

Amino acid position 37 of HLA-DRß1 affects susceptibility to Crohn's disease in Asians.

Crohn's disease (CD) and ulcerative colitis (UC) are the major types of chronic inflammatory bowel disease (IBD) characterized by recurring episodes of inflammation of the gastrointestinal tract. Although it is well established that human leukocyte antigen (HLA) is a major risk factor for IBD, it is yet to be determined which HLA alleles or amino acids drive the risks of CD and UC in Asians. To define the roles of HLA for IBD in Asians, we fine-mapped HLA in 12 568 individuals from Korea and Japan (3294 patients with CD, 1522 patients with UC and 7752 controls). We identified that the amino acid position 37 of HLA-DRß1 plays a key role in the susceptibility to CD (presence of serine being protective, P = 3.6 × 10-67, OR = 0.48 [0.45-0.52]). For UC, we confirmed the known association of the haplotype spanning HLA-C*12:02, HLA-B*52:01 and HLA-DRB1*1502 (P = 1.2 × 10-28, OR = 4.01 [3.14-5.12]).

Preliminary study on alginate/NIPAM hydrogel-based soft microrobot for controlled drug delivery using electromagnetic actuation and near-infrared stimulus.

Currently, microrobots are receiving attention because of their small size and motility, which can be applied to minimal invasive therapy. Additionally, various microrobots using hydrogel with the characteristics of biocompatibility and biodegradability are also being developed. Among them, microrobots that swell and deswell in response to temperature changes caused by external near infrared (NIR) stimuli, focused ultrasound, and an alternating magnetic field, have been receiving a great amount of interest as drug carriers for therapeutic cell delivery. In this study, we propose a spring type medical microrobot that can be manipulated by an electromagnetic actuation (EMA) system and respond to an external stimulus (NIR). Additionally, we verified its feasibility with regard to targeting and drug delivery. There exist various methods of fabricating a spring type microrobot. In this study, we adopted a simple method that entails using a perfluoroalkoxy (PFA) microtube and a syringe pump. Moreover, we also used a hydrogel mixture composed of natural alginate, N-Isopropylacrylamide (NIPAM) for temperature responsiveness, and magnetic nanoparticles (MNPs) for electromagnetic control. Then, we fabricated a spring type alginate/NIPAM hydrogel-based soft microrobot. Additionally, we encapsulated doxorubicin (DOX) for tumor therapy in the microrobot. To verify the feasibility of the proposed spring type hydrogel-based soft microrobot's targeting and drug delivery, we developed an EMA and NIR integrated system. Finally, we observed the swelling and deswelling of the soft microrobot under NIR stimulation and verified the EMA controlled targeting. Moreover, we implemented a control function to release the encapsulated anticancer drug (DOX) through the swelling and deswelling of the soft microrobot by NIR, and evaluated the feasibility of cancer cell therapy by controlling the release of the drug from the soft microrobot.

Efficient micro-cavity top emission OLED with optimized Mg:Ag ratio cathode.

Micro-cavity top-emitting organic light emitting diodes (TEOLEDs) are now receiving prominence as a technology for the active matrix display applications. The semi-transparent metal cathode plays the crucial role in realizing TEOLEDs structure. Here, we report the optimization results on Mg:Ag ratio as the semitransparent cathode deposited by vacuum thermal evaporation. The optimized Mg:Ag cathode with 1:10 ratio (wt %) shows a sheet resistance value as low as 5.2 Ω/□, an average transmittance of 49.7%, reflectance of 41.4%, and absorbance of 8.9% over the visible spectral region (400~700 nm). The fabricated red TEOLEDs device implemented using LiF (1nm)/Mg:Ag (1:10) cathode shows the voltage value of 4.17 V at a current density of 10.00 mA/cm2, and current efficiencies variation from 55.3 to 50.1 cd/A over the brightness range 2,000 - 12,000 cd/m2. The electroluminescence (EL) spectrum displays the light emission at 608 nm wavelength with a half width of 29.5 nm. The narrow half-width of red light emission is attributed to the micro-cavity effects due to the semitransparent cathode.

Automated acquisition and analysis of airway surface liquid height by confocal microscopy.

The airway surface liquid (ASL) is a thin-liquid layer that lines the luminal side of airway epithelia. ASL contains many molecules that are involved in primary innate defense in the lung. Measurement of ASL height on primary airway cultures by confocal microscopy is a powerful tool that has enabled researchers to study ASL physiology and pharmacology. Previously, ASL image acquisition and analysis were performed manually. However, this process is time and labor intensive. To increase the throughput, we have developed an automatic ASL measurement technique that combines a fully automated confocal microscope with novel automatic image analysis software that was written with image processing techniques derived from the computer science field. We were able to acquire XZ ASL images at the rate of ∼ 1 image/s in a reproducible fashion. Our automatic analysis software was able to analyze images at the rate of ∼ 32 ms/image. As proofs of concept, we generated a time course for ASL absorption and a dose response in the presence of SPLUNC1, a known epithelial sodium channel inhibitor, on human bronchial epithelial cultures. Using this approach, we determined the IC50 for SPLUNC1 to be 6.53 µM. Furthermore, our technique successfully detected a difference in ASL height between normal and cystic fibrosis (CF) human bronchial epithelial cultures and detected changes in ATP-stimulated Cl(-)/ASL secretion. We conclude that our automatic ASL measurement technique can be applied for repeated ASL height measurements with high accuracy and consistency and increased throughput.

A hybrid actuated microrobot using an electromagnetic field and flagellated bacteria for tumor-targeting therapy.

In this paper, we propose a new concept for a hybrid actuated microrobot for tumor-targeting therapy. For drug delivery in tumor therapy, various electromagnetic actuated microrobot systems have been studied. In addition, bacteria-based microrobot (so-called bacteriobot), which use tumor targeting and the therapeutic function of the bacteria, has also been proposed for solid tumor therapy. Compared with bacteriobot, electromagnetic actuated microrobot has larger driving force and locomotive controllability due to their position recognition and magnetic field control. However, because electromagnetic actuated microrobot does not have self-tumor targeting, they need to be controlled by an external magnetic field. In contrast, the bacteriobot uses tumor targeting and the bacteria's own motility, and can exhibit self-targeting performance at solid tumors. However, because the propulsion forces of the bacteria are too small, it is very difficult for bacteriobot to track a tumor in a vessel with a large bloodstream. Therefore, we propose a hybrid actuated microrobot combined with electromagnetic actuation in large blood vessels with a macro range and bacterial actuation in small vessels with a micro range. In addition, the proposed microrobot consists of biodegradable and biocompatible microbeads in which the drugs and magnetic particles can be encapsulated; the bacteria can be attached to the surface of the microbeads and propel the microrobot. We carried out macro-manipulation of the hybrid actuated microrobot along a desired path through electromagnetic field control and the micro-manipulation of the hybrid actuated microrobot toward a chemical attractant through the chemotaxis of the bacteria. For the validation of the hybrid actuation of the microrobot, we fabricated a hydrogel microfluidic channel that can generate a chemical gradient. Finally, we evaluated the motility performance of the hybrid actuated microrobot in the hydrogel microfluidic channel. We expect that the hybrid actuated microrobot will be utilized for tumor targeting and therapy in future.

Learning-Based Ordering Characters on Ancient Document.

Digitalizing and translating a scanned document image entails detecting the characters using a detector and translating the characters in the order they were detected with a translator. However, it is impossible to translate these characters correctly because the detector often detects them in any order. As a result, since it is critical to organize the recognized characters for proper translation, we propose ordering characters from documents with multiple variations using the strength of the learning-based model that learns the necessary operations from the data. In this task, it is difficult to order the characters written on antique handwritten documents that have deviations such as a bent or split line, as opposed to official records that have lines placed uprightly one by one. Because dealing with these many variants using a human-designed algorithm is problematic, we arrange characters printed on papers with diverse variations by taking advantage of a training model that can learn the appropriate function from data. Our method outputs both line id and y-axis and combines them to assign the sequential index. It is difficult to train using simply local regions because sequential character indexes in a large range include long-range dependencies. To solve this problem, we use network architecture to expand the receptive field as wide as possible. The network must learn to give various indexes to characters in similar places for each document because the number and area of characters vary for each document. We offer the ground truth assign method based on the absolute position to assign similar indexes to characters in similar places. Furthermore, even if the network uses absolute ground truth, the network may assign the incorrect line if the center coordinates of characters are biased in one direction. As a result, we employed the Region of Interest (ROI) from the pretrained coordinate layer, which contains position and trend information. We used the modified edit distance to compare the similarity of character indexes from the ground truth and our technique. In addition, we computed the modified fisher criterion to assess the degree of the clustering line. Consequently, our edit distance is just 0.43 times that of the human-designed algorithm, and our fisher criterion is 1.46 times that of the human-designed algorithm, improving the performance of human-designed algorithm.

Optimal Energetic-Trap Distribution of Nano-Scaled Charge Trap Nitride for Wider Vth Window in 3D NAND Flash Using a Machine-Learning Method.

A machine-learning (ML) technique was used to optimize the energetic-trap distributions of nano-scaled charge trap nitride (CTN) in 3D NAND Flash to widen the threshold voltage (Vth) window, which is crucial for NAND operation. The energetic-trap distribution is a critical material property of the CTN that affects the Vth window between the erase and program Vth. An artificial neural network (ANN) was used to model the relationship between the energetic-trap distributions as an input parameter and the Vth window as an output parameter. A well-trained ANN was used with the gradient-descent method to determine the specific inputs that maximize the outputs. The trap densities (NTD and NTA) and their standard deviations (σTD and σTA) were found to most strongly impact the Vth window. As they increased, the Vth window increased because of the availability of a larger number of trap sites. Finally, when the ML-optimized energetic-trap distributions were simulated, the Vth window increased by 49% compared with the experimental value under the same bias condition. Therefore, the developed ML technique can be applied to optimize cell transistor processes by determining the material properties of the CTN in 3D NAND Flash.

Uncorrelated feature encoding for faster image style transfer.

Recent image style transfer methods use a pre-trained convolutional neural network as their feature encoder. However, the pre-trained network is not optimal for image style transfer but rather for image classification. Furthermore, they require time-consuming feature alignment to consider the existing correlation among channels of the encoded feature map. In this paper, we propose an end-to-end learning method that optimizes both encoder and decoder networks for style transfer task and relieves the computational complexity of the existing correlation-aware feature alignment. First, we performed end-to-end learning that updates not only decoder but also encoder parameters for the task of image style transfer in the network training phase. Second, in addition to the previous style and content losses, we use uncorrelation loss, i.e., the total correlation coefficient among responses of encoder channels. Our uncorrelation loss allows the encoder network to generate a feature map of channels without correlation. Subsequently, our method results in faster forward processing with only a light-weighted transformer of correlation-unaware feature alignment. Moreover, our method drastically reduced the channel redundancy of the encoded feature during the network training process. This provides us a possibility to perform channel elimination with negligible degradation in generated style quality. Our method is applicable to multiple scaled style transfer by using the cascade network scheme and allows a user to control style strength through the usage of a content-style trade-off parameter.

Atomic-Layer-Deposited SiOx/SnOx Nanolaminate Structure for Moisture and Hydrogen Gas Diffusion Barriers.

High-density SnOx and SiOx thin films were deposited via atomic layer deposition (ALD) at low temperatures (100 °C) using tetrakis(dimethylamino)tin(IV) (TDMASn) and di-isopropylaminosilane (DIPAS) as precursors and hydrogen peroxide (H2O2) and O2 plasma as reactants, respectively. The thin-film encapsulation (TFE) properties of SnOx and SiOx were demonstrated with thickness dependence measurements of the water vapor transmission rate (WVTR) evaluated at 50 °C and 90% relative humidity, and different TFE performance tendencies were observed between thermal and plasma ALD SnOx. The film density, crystallinity, and pinholes formed in the SnOx film appeared to be closely related to the diffusion barrier properties of the film. Based on the above results, a nanolaminate (NL) structure consisting of SiOx and SnOx deposited using plasma-enhanced ALD was measured using WVTR (H2O molecule diffusion) at 2.43 × 10-5 g/m2 day with a 10/10 nm NL structure and time-lag gas permeation measurement (H2 gas diffusion) for applications as passivation layers in various electronic devices.

Simultaneous emission of orthogonal handedness in circular polarization from a single luminophore.

The direct emission of circularly polarized (CP) light improves the efficiency of an organic light-emitting diode and characterizes the secondary structure of proteins. In most cases, CP light is generated from a luminescent layer containing chiral characteristics, thereby generating only one kind of CP light in an entire device. Here, we propose direct CP light emissions using a twisted achiral conjugate polymer without any chiral dopant as an emitting layer (EML). The twisted structure is induced in the mesogenic conjugate polymer due to its elasticity by applying different alignment directions to its upper and lower interfaces. Furthermore, we demonstrate the simultaneous emission of orthogonal CP light in a single luminescent device by patterning different alignment directions on the surfaces of the EML. The light source with multipolarization including the orthogonal CP states is applicable to many applications in biosensors and optical devices.

Degradation of OLED performance by exposure to UV irradiation.

Organic light-emitting diode (OLED) displays are highly susceptible to the harsh environmental conditions found outdoors, like exposure to direct sunlight as well as UV radiation and storage temperature, resulting in a loss of luminance and lifespan, pixel shrinkage, and permanent damage and/or malfunction of the panel. Here, we fabricated top emission OLEDs (TEOLEDs) using Yb : LiF (1 : 1, 2 nm)/Ag : Mg (10 : 1, 16 nm) and Mg : LiF (1 : 1, 2 nm)/Ag : Mg (10 : 1, 16 nm) cathode units and the performances of the devices were investigated by subjecting them to UV radiation. A fabricated red TEOLED (control device), employing a standard Mg : LiF (1 : 1, 2 nm) electron injection layer (EIL) and an Ag : Mg (16 nm) cathode, showed a rapid decrease in luminance and a fast increase in driving voltage at 10 mA cm-2 over time after UV irradiation for 300 h. However, a cathode unit comprising a Yb : LiF (1 : 1, 2 nm) EIL and an Ag : Mg (10 : 1, 16 nm) cathode showed no loss of luminance or increase in driving voltage at 10 mA cm-2 over time after UV irradiation for 300 h. Therefore, we investigated the changes occurring in both cathode units due to UV irradiation using the lift-out FIB-TEM technique and EDS mapping. With UV irradiation for 300 h, Ag atoms migrated toward the center of the cathode, Mg atoms migrated toward the CPL, and no Mg atoms were observed in the EIL area. In contrast, we observed (i) no substantial migration of Ag atoms and they were located at the center of the cathode, (ii) no migration of Mg atoms toward the CPL layer, and (iii) no movement of Yb atoms after UV irradiation. Furthermore, the UV irradiated red TEOLED with an Mg : LiF (1 : 1, 2 nm) EIL showed (i) deterioration in electron injection into the emissive layer (EML) and an increase in the EIL/metal interface resistance, and (ii) a remarkable shift of the J-V curve to the higher voltage side, while almost no such changes were observed in the TEOLD with a Yb : LiF (1 : 1, 2 nm) EIL. Also, an almost identical RGB pixel emitting area was noticed in the Yb : LiF (1 : 1, 2 nm) based devices after UV irradiation for 300 h. These results suggest that Yb could become a good candidate for the cathode unit, providing better device stability against harsh environmental conditions as well as excellent electron injection properties.

Nonparametric Facial Feature Localization Using Segment-Based Eigenfeatures.

We present a nonparametric facial feature localization method using relative directional information between regularly sampled image segments and facial feature points. Instead of using any iterative parameter optimization technique or search algorithm, our method finds the location of facial feature points by using a weighted concentration of the directional vectors originating from the image segments pointing to the expected facial feature positions. Each directional vector is calculated by linear combination of eigendirectional vectors which are obtained by a principal component analysis of training facial segments in feature space of histogram of oriented gradient (HOG). Our method finds facial feature points very fast and accurately, since it utilizes statistical reasoning from all the training data without need to extract local patterns at the estimated positions of facial features, any iterative parameter optimization algorithm, and any search algorithm. In addition, we can reduce the storage size for the trained model by controlling the energy preserving level of HOG pattern space.