Exploring Distance-Aware Weighting Strategies for Accurate Reconstruction of Voxel-Based 3D Synthetic Models

Abstract

Abstract. In this paper, we propose and evaluate various distanceaware weighting strategies to improve reconstruction accuracy of a voxelbased model according to the Truncated Signed Distance Function (TSDF), from the data obtained by low-cost depth sensors. We look at two strategy directions: (a) weight definition strategies prioritizing importance of the sensed data depending on the data accuracy, and (b) model updating strategies defining the level of influence of the new data on the existing 3D model. In particular, we introduce Distance-Aware (DA) and Distance-Aware Slow-Saturation (DASS) updating methods to intelligently integrate the depth data into the synthetic 3D model based on the distance-sensitivity metric of a low-cost depth sensor. By quantitative and qualitative comparison of the resulting synthetic 3D models to the corresponding ground-truth models, we identify the most promising strategies, which lead to an accuracy improvement involving a reduction of the model error by 10 − 35%.

Evaluation of Distance-Aware KinFu Algorithm for Stereo Outdoor Data

“…To eliminate this constraint, the DASS method has been introduced as below (Javan Hemmat et al, 2014a;Javan Hemmat et al, 2014b).…”

“…In this paper, we utilize the weight definition which has been already proposed in (Javan Hemmat et al, 2014a;Javan Hemmat et al, 2014b) to guarantee the assignment of higher weights to points in the scene located on closer distances to the sensor. Based on the sensor features and scene characteristics, there is a valid range for depth data, defined between a maximum and a minimum distance, d max and d min , respectively.…”

“…While recent evaluation of the Kinect intrinsics has revealed relative robustness to ambient light, incidence angle, and radiometric influences, the important sensor limitation is the low accuracy for large distance measurements (Chow et al, 2012;Khoshelham, 2011;Khoshelham and Elberink, 2012). Since the default weighting strategy of the TSDF model is not capable to preserve more accurate data against less accurate data, an improved approach has been proposed by introduction of weight definitions and updating strategies (Javan Hemmat et al, 2014a). The resulting improvements in model quality (Javan Hemmat et al, 2014a) and pose-estimation accuracy (Javan Hemmat et al, 2014b) obtained by these strategies, have motivated us to extend the approach to the domain of stereo sensor data.…”

“…Since the default weighting strategy of the TSDF model is not capable to preserve more accurate data against less accurate data, an improved approach has been proposed by introduction of weight definitions and updating strategies (Javan Hemmat et al, 2014a). The resulting improvements in model quality (Javan Hemmat et al, 2014a) and pose-estimation accuracy (Javan Hemmat et al, 2014b) obtained by these strategies, have motivated us to extend the approach to the domain of stereo sensor data.…”

Real-time planar segmentation of depth images: from three-dimensional edges to segmented planes

“…3(a). The threshold can have fixed setting or alternatively be a function that returns a corresponding threshold value according to the depth value of a point (distance-aware threshold 28,29 ). The unit for Th jump is consistent with the unit utilized for depth images.…”