Image Selection Process Research Articles

Image Selection Method from Image Sequence to Improve Computational Efficiency of 3D Reconstruction: Application of Robust Threshold Based on Multimodal Test to Images

This paper describes the image selection method by multimodal detection for improving the computational efficiency of three-dimensional (3D) reconstruction based on images in a time-series. To decrease the calculation time of the 3D reconstruction, an adequate selection from the images is required. For this reason, we introduced multimodal detection by a statistical test on the image selection process, and then applied it to soundness evaluation of the displacements based on the optical flow between images acquired by a camera. The results suggest that suitable images can be extracted from the images in a time-series for decreasing the calculation time of the 3D reconstruction. Therefore, the suitable images selected by the proposed method contributed to efficiently performing the 3D reconstruction.

복합문화공간 이미지 평가에 나타난 성별 감성특성

This study conducted the sensibility evaluation survey by using 3 perception types and 21 adjective pairs to targetting 4 images. The study results that analyzed visual perception characteristics in gender can be defined in the following several ways. First, it was possible to sort out the dominant characteristics by gender, perception types in order through comparing the average of sensibility evaluation values and the proportion of images that were selected as the most dominant ratio. This has differentiated compares to the existing study and can be an analysis technique that can compare more specific images and sensibility characteristics by analyzing gender sensibility characteristics toward images. Second, in the spatial evaluation, females had a clear selection of space than males and also showed higher results in the sensibility evaluation as well as indicated characteristics to express clearly. Third, through in the number of adjective pairs common to gender by perception types, 12 adjective pairs showed more than 10% advantage in spatiality and of which 10 adjective pairs showed gender congruence. 8 adjective pairs showed advantages in the formality and all these 8 adjective pairs indicated as common in both genders, which can be a characteristic of common gender perception towards given images. There were 5 adjective pairs shown as advantages in decorativeness, of which 1 adjective pair resulted as gender congruence. This result can be seen that decorativeness is the lowest gender commonality and lowest frequency of selection from images. Fourth, it was able to confirm that gender prefers to make a different selection depends on space or adjective pairs. It was possible to clarify the differences in the gender sensibility characteristics which appears in each perception types/ adjective pairs by arranging the big difference values of sensibility adjectives that appeared in the image selection process by gender. Depends on the gender, there were the characteristics that the selection of each image towards the perception types was clearly different, and furthermore, it is able to expect that this kind of space will bring more clear characteristics by the gender.

INTRA- AND INTER-OBSERVER AGREEMENT OF PROXIMAL HUMERAL FRACTURES CLASSIFICATIONS IN ADULTS.

Evaluating intra- and inter-observer agreement of the Neer, AO, and AO/OTA proximal humerus fractures classification systems in adults. In total, 100 X-rays of patients with proximal humerus fractures were selected according to the inclusion and exclusion criteria established in this study. They were evaluated by four evaluators with different levels of expertise. The evaluation was performed at two distinct moments, with an interval of 21 days between each analysis. Images were randomized for the second evaluation by a researcher who did not participate in the image selection process. A Fleiss Kappa test was performed to evaluate intra- and inter-observer agreement. We observed a substantial agreement with k = 0.669, k = 0.715, and k = 0.780 for the Neer, AO, and AO/OTA classification systems, respectively. In the second evaluation, intra-observer agreement improved. In the first evaluation, we obtained values of k = 0.724, k = 0.490, and k = 0.599 for the evaluation of the Neer, AO, and AO/OTA classifications. In the second evaluation, the values k = 0.759, k = 0.772, and k = 0.858. Therefore, the evaluations went from moderate to substantial for the AO classification and from moderate to practically perfect for the AO/OTA classification. The level of inter-observer agreement was substantial (0.61-0.80), with k = 0.669, k = 0.715, and k = 0.780 for the Neer, AO, and AO/OTA classifications, respectively. Level of Evidence III, Cross-Sectional Observational Study.

GEO-TAGGED IMAGE RETRIEVAL FROM MAPILLARY STREET IMAGES FOR A TARGET BUILDING

Abstract. This study aims to investigate the possibility to automate the image selection process for the target building from Mapillary images through a web application where the user only initiates one image of the target building as a query. Using the data provided with Mapillary API and Overpass API, all images having full or partial coverage of the target building were selected. Then the images were segmented by using a pre-trained U-Net model to discard any images having less than 20% building coverage. The experiments showed promising results yielding 0.971 and 0.887 of overall accuracy after segmentation steps for two different target buildings.

Improving the calculation accuracy of FEM for asphalt mixtures in simulation of SCB test considering the mesostructure characteristics

ABSTRACT Fracture simulation of asphalt mixtures based on three-dimensional finite element (3D-FE) simulation considering the mesostructure consumes too much computing resources and has low efficiency, which limits the popularisation and application of asphalt mixture fracture characteristics simulations. In contrast, two-dimensional finite element (2D-FE) simulation needs less computational resources and has high efficiency. However, 2D-FE simulation results significantly impacted by aggregate content and particle distribution in the scanned images and an arbitrary 2D image could hardly reflect the mesostructure of the whole asphalt mixture specimen. Therefore, two internal structure indicators, such as aggregate content indicator (ACI) and aggregate distribution indicator (ADI), were defined for image assessment, and an image selection process was proposed based on these two indicators for 2D-FE modelling method. By comparing the results obtained from 3D-FE simulations, 2D-FE simulations with or without image selection process, and laboratory tests, it can be found that the results of 2D-FE simulations with the image selection process are more consistent with laboratory test than other two simulation processes. Thus, it can be verified that the image selection process can highly enhance the accuracy of 2D-FE simulation while reducing the number of models needed and improving computational efficiency.

A systematic evaluation of influence of image selection process on remote sensing-based burn severity indices in North American boreal forest and tundra ecosystems

Satellite imagery has been widely used for the assessment of wildfire burn severity within the scientific community and fire management agencies. Multiple indices have been proposed to assess burn severity, among which the differenced Normalized Burn Ratio (dNBR) is arguably the most commonly used index that is expected to provide an objective and consistent assessment. However, although evidence of variability in the dNBR-based assessment of burn severity driven by image pair selection has been shown in many studies, the comprehensive examination of the extent of the bias resulting from the image selection has been lacking. In this study, we focus on three factors of the image selection process which are encountered by most Landsat-derived dNBR applications, including the sensor combination and the difference in timing of image acquisition (for both the year and seasonality) of pre- and post-fire image pairs. Through separate analyses, each targeting a single factor, we show that Landsat sensor combination between the pre- and post-fire images has a limited impact on the dNBR values. The difference in the year of acquisition between the images in the image pairs is shown to influence dNBR assessment with a noticeable increase in mean dNBR (>0.1) with only a single year difference between images compared to multi-year differences. However, differences in the image acquisition seasons and the resulting phenological differences is shown to impact dNBR values most considerably. Based on our results, we warn against the calculation of dNBR when the images are acquired in different seasons. We believe that despite the existence of multiple derivatives of dNBR, there remains a need for an improved version; one that is less susceptible to the phenological impacts introduced by the selected images.

Breast image pre-processing for mammographic tissue segmentation

During mammographic image acquisition, a compression paddle is used to even the breast thickness in order to obtain optimal image quality. Clinical observation has indicated that some mammograms may exhibit abrupt intensity change and low visibility of tissue structures in the breast peripheral areas. Such appearance discrepancies can affect image interpretation and may not be desirable for computer aided mammography, leading to incorrect diagnosis and/or detection which can have a negative impact on sensitivity and specificity of screening mammography. This paper describes a novel mammographic image pre-processing method to improve image quality for analysis. An image selection process is incorporated to better target problematic images. The processed images show improved mammographic appearances not only in the breast periphery but also across the mammograms. Mammographic segmentation and risk/density classification were performed to facilitate a quantitative and qualitative evaluation. When using the processed images, the results indicated more anatomically correct segmentation in tissue specific areas, and subsequently better classification accuracies were achieved. Visual assessments were conducted in a clinical environment to determine the quality of the processed images and the resultant segmentation. The developed method has shown promising results. It is expected to be useful in early breast cancer detection, risk-stratified screening, and aiding radiologists in the process of decision making prior to surgery and/or treatment.

Differences in facial profile and dental esthetic perceptions between young adults and orthodontists

The principal aim of this study was to investigate differences in perception of soft-tissue facial profiles and dental esthetics between young Chinese adults and orthodontists. Eight hundred ninety-two subjects (444 male, 448 female), ages 16 to 24 years, chose 1 image from among 5 profile silhouettes and from among 10 ranked color photographs of the aesthetic component (AC) of the index of orthodontic treatment need that most closely resembled their own profile and dental esthetic appearance, respectively. A panel of 20 orthodontists then independently repeated the same image selection process. Each subject also completed the Eysenck personality questionnaire for psychoticism. We used the Mann-Whitney U test and the Spearman rank correlation test, with statistical significance set at α= 0.05. Only 37.0% of subjects had straight profiles by objective orthodontic assessment, but 85.0% chose straight profiles by subjective self-perception. About 17.5%, mainly females, chose the mild convex as the ideal profile. Only 2.5% of the subjects were ranked 1 on the AC by orthodontists, but 43.6% chose 1, or ideal, by self-perception. Male subjects scored significantly higher than did female subjects for self-perceived facial profiles (more protruded chins) and for the AC (more attractive dental appearance). Subjects with high psychoticism Tscores (>50) scored significantly lower for self-perceived facial profiles (more retruded chins) and on the AC (less attractive dental appearance). Young Chinese adults perceived their facial profiles and dental appearances to be significantly more straight and attractive, respectively, than did the orthodontists. A significant proportion of the young adults, mainly women, preferred a mild convex facial profile. High psychoticism scores might significantly affect the self-perception of orthodontic treatment needs.

Enhanced SAR-Based Snow-Covered Area Estimation Method for Boreal Forest Zone

In this paper, an enhanced method for fractional snow-covered area (SCA) estimation for the boreal forest zone is presented. The new approach, based on utilizing weather station data alongside with spaceborne synthetic aperture radar (SAR) imagery, leads to a significantly improved estimation accuracy. While the Helsinki University of Technology (TKK) SAR-based SCA estimation method serves as a basic tool in the SCA estimation, the ground-based weather station observations are employed to still strengthen its performance at the nearly melt-off or totally melt-off conditions. The method is still improved by a new reference image selection process, leading to more accurate results and an easier adaptivity to new areas. The SCA estimation accuracy of the new enhanced method is compared with optical satellite-based SCA data. Evaluation of the method is carried out using Radarsat wide-swath data for the snow-melt seasons of 2004-2006. The results show a significant increase in accuracy when the enhanced SCA method is applied. Correlation between the radar-based and optical comparison data increases from 0.914 to 0.947 and root-mean-square error improves from 0.151 to 0.123 with the new method. Traditionally, the TKK method has provided SCA estimates for Finnish third-order subdrainage basins. In this paper, the method is adapted to produce SCA estimates also in 5 times 5 km spatial resolution. The analyses for the 5 times 5 km method indicate poorer estimation accuracy than the nominal drainage-basin-based method.

Stochastic exploration and active learning for image retrieval

This paper deals with content-based image retrieval. When the user is looking for large categories, statistical classification techniques are efficient as soon as the training set is large enough. We introduce a two-step—exploration, classification—interactive strategy designed for category retrieval. The first step aims at getting a useful initial training set for the classification step. A stochastic image selection process is used instead of the usual strategy based on a similarity score ranking. This process is dedicated to explore the database in order to collect examples as various as possible of the searched category. The second step aims at providing the best classification between relevant and irrelevant images. Based on SVM, the classification applies an active learning strategy through user interaction. A quality assessment is carried out on the ANN and COREL databases in order to compare and validate our approach.