Clustering-Based Plane Segmentation Neural Network for Urban Scene Modeling

Urban scene modeling is a challenging but essential task for various applications, such as 3D map generation, city digitization, and AR/VR/metaverse applications. To model man-made structures, such as roads and buildings, which are the major components in general urban scenes, we present a clustering-based plane segmentation neural network using 3D point clouds, called hybrid K-means plane segmentation (HKPS). The proposed method segments unorganized 3D point clouds into planes by training the neural network to estimate the appropriate number of planes in the point cloud based on hybrid K-means clustering. We consider both the Euclidean distance and cosine distance to cluster nearby points in the same direction for better plane segmentation results. Our network does not require any labeled information for training. We evaluated the proposed method using the Virtual KITTI dataset and showed that our method outperforms conventional methods in plane segmentation. Our code is publicly available.

A Change of Scenery: Does Exposure to Images of Nature Affect Delay Discounting and Food Desirability?

Previous research suggests that exposure to nature may reduce delay discounting (the tendency to discount larger future gains in favor of smaller immediate rewards) and thereby facilitate healthier dietary intake. This pre-registered study examined the impact of online exposure to images of natural scenes on delay discounting and food preferences. It was predicted that exposure to images of natural scenes (vs. images of urban scenes) would be associated with: (i) lower delay discounting; (ii) higher desirability for fruits and vegetables (and lower desirability for more energy-dense foods); and (iii) delay discounting would mediate the effect of nature-image exposure on food desirability. Adult participants (N = 109) were recruited to an online between-subjects experiment in which they viewed a timed sequence of six images either showing natural landscape scenes or urban scenes. They then completed measures of mood, delay discounting (using a five-trial hypothetical monetary discounting task) and rated their momentary desire to eat four fruits and vegetables (F&V), and four energy-dense foods. There was no statistically significant effect of experimental condition (natural vs. urban image exposure) on delay discounting or food desirability. Bayes factors supported the null hypothesis for discounting (BF01 = 4.89), and energy-dense food desirability (BF01 = 7.21), but provided no strong evidence for either hypothesis for F&V desirability (BF01 = 0.78). These findings indicate that brief online exposure to images of nature does not affect momentary impulsivity or energy-dense food preference, whereas for preference for less-energy dense foods, the evidence was inconclusive.

Street-Level Image Localization Based on Building-Aware Features via Patch-Region Retrieval under Metropolitan-Scale

The aim of image-based localization (IBL) is to localize the real location of query image by matching reference image in database with GNSS-tags. Popular methods related to IBL commonly use street-level images, which have high value in practical application. Using street-level image to tackle IBL task has the primary challenges: existing works have not made targeted optimization for urban IBL tasks. Besides, the matching result is over-reliant on the quality of image features. Methods should address their practicality and robustness in engineering application, under metropolitan-scale. In response to these, this paper made following contributions: firstly, given the critical of buildings in distinguishing urban scenes, we contribute a feature called Building-Aware Feature (BAF). Secondly, in view of negative influence of complex urban scenes in retrieval process, we propose a retrieval method called Patch-Region Retrieval (PRR). To prove the effectiveness of BAF and PRR, we established an image-based localization experimental framework. Experiments prove that BAF can retain the feature points that fall on the building, and selectively lessen the feature points that fall on other things. While this effectively compresses the storage amount of feature index, we can also improve recall of localization results; implemented in the stage of geometric verification, PRR compares matching results of regional features and selects the best ranking as final result. PRR can enhance effectiveness of patch-regional feature. In addition, we fully confirmed the superiority of our proposed methods through a metropolitan-scale street-level image dataset.

Development of Smart 3D Object Detection using LiDAR Cloud-Data Collection

Smart cities are made up of autonomous vehicle and they communicate and interact with their environment and require high precision computer vision to maintain driver and pedestrian safety. This paper presents a cost-efficient, non-intrusive and easy to use method for collecting data traffic counts using LiDAR technology. The proposed method incorporates a LiDAR sensor, a Convolutional Neural Network (CNN) and a Hybrid SVM into a single traffic counting framework. As the technology is economical and readily accessible, LiDAR is adopted. The distance data obtained are translated into the signals. Due to the difficulty of urban scenes, automatic detection of objects from remotely sensed data within urban areas is difficult. While recent advances in computer vision have shown that CNNs are very suitable for this task, the design and training of CNNs of this kind remained demanding and time consuming, given the challenge of collecting a large and well-annotated dataset and the specificity of every task. Hybrid SVM is a supervised data classification and regression machine learning tool. In the methodology the Hybrid SVM is used in detection and non-detection cases of highly complex distance data points obtained from the sensor. In order to examine the performance of the proposed method, the test is carried out in three different locations in Alexandria, Egypt. The results of tests show that the pro-imposed method achieves acceptable results in vehicle collection, which results in a precision of 85–89%. The exactness of the method proposed is determined by the colour of a vehicle’s external surface.

Gaze correlates of view preference: Comparing natural and urban scenes

When looking out of a window, natural views are usually associated with restorative qualities and are given a higher preference than urban scenes. Previous research has shown that gaze behaviour might differ based on the natural or urban content of views. A lower number of fixations has been associated with the aesthetic evaluation of natural scenes while, when looking at an urban environment, a high preference has been correlated with more exploratory gaze behaviours. To characterise gaze correlates of view preference across natural and urban scenes, we collected and analysed experimental data featuring subjective preference ratings, eye-tracking measures, verbal reasoning associated with preference and nature relatedness scores. Consistent with the literature, our results confirm that natural scenes are more preferred than urban views and that gaze behaviours depend on view type and preference. Observing natural scenes was characterised by lower numbers of fixations and saccades, and longer fixation durations, compared to urban views. However, for both view types, most preferred scenes led to more fixations and saccades. Our findings also showed that nature relatedness may be correlated with visual exploration of scenes. Individual preferences and personality attributes, therefore, should be accounted for in studies on view preference and gaze behaviour.

Dynamic multi-sensor platform for efficient 3D-digitalization of cities: part II

The digitization of geographic environments, such as cities and archaeological sites, is of priority interest to the scientific community due to its potential applications. But there are still several issues to address. There are various digitization strategies, which include terrestrial/ airborne platforms and composed of various sensors, among the most common, cameras and laser scanners. A comprehensive methodology is presented to reconstruct urban environments using a mobile land platform. All the implemented stages are described, which includes the acquisition, processing, and correlation of the data delivered by a Velodyne HDL-64E scanner, a spherical camera, GPS, and inertial systems. The process to merge several point clouds to build a large-scale map is described, as well as the generation of surfaces. Being able to render large urban areas using a low density of points but without losing the details of the structures within the urban scenes. The proposal is evaluated using several metrics, for example, Coverage and Root-Mean-Square-Error (RSME). The results are compared against 3 methodologies reported in the literature. Obtaining better results in the 2D/3D data fusion process and the generation of surfaces. The described method has a low RMSE (0.79) compared to the other methods and a runtime of approximately 40 seconds to process each data set (point cloud, panoramic image, and inertial data). In general, the proposed methodology shows a more homogeneous density distribution without losing the details, that is, it conserves the spatial distribution of the points, but with fewer data.