The Comparison of Keypoint Detectors and Descriptors for Registration of RGB-D Data

Kostusiak, Aleksander

doi:10.1007/978-3-319-29357-8_53

Cited by 8 publications

(5 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This process is repeated for all incoming RGB-D data frames. Then, point features from the newest frame are matched to the ones from the previous frame with cross-checking of the matchings to clear out spurious associations as soon as possible [18].…”

Section: System Structurementioning

confidence: 99%

“…We use RANSAC [13] procedure twice to remove the remaining bad-matches while transformations between the consecutive frames are computed [18]. In each of the RANSAC instances, we draw three point pairs, which is the minimal number of points required by the Kabsch algorithm [15] to determine the rototranslation between two successive frames.…”

Section: System Structurementioning

confidence: 99%

“…Motivated by the results of previous experiments [18], we use the AKAZE algorithm [1] for feature detection and description. This detector-descriptor al-lows to achieve point feature detection and matching results that are very similar to the popular SURF detector-descriptor performance, but with the use of a binary descriptor, which requires less computation power in the matching process.…”

Section: System Structurementioning

confidence: 99%

See 2 more Smart Citations

On the Efficiency of Population-Based Optimization in Finding Best Parameters for RGB-D Visual Odometry

Kostusiak¹,

Skrzypczyński²

2019

JAMRIS

Self Cite

View full text Add to dashboard Cite

Visual odometry es mates the transforma ons between consecu ve frames of a video stream in order to recover the camera's trajectory. As this approach does not require to build a map of the observed environment, it is fast and simple to implement. In the last decade RGB-D cameras proliferated in robo cs, being also the sensors of choice for many prac cal visual odometry systems. Although RGB-D cameras provide readily available depth images, that greatly simplify the frame-to-frame transforma ons computa on, the number of numerical parameters that have to be set properly in a visual odometry system to obtain an accurate trajectory es mate remains high. Whereas se ng them by hand is certainly possible, it is a tedious try-and-error task. Therefore, in this ar cle we make an assessment of two popula on-based approaches to parameter op miza on, that are for long me applied in various areas of robo cs, as means to find best parameters of a simple RGB-D visual odometry system. The op miza on algorithms inves gated here are parcle swarm op miza on and an evolu onary algorithm variant. We focus on the op miza on methods themselves, rather than on the visual odometry algorithm, seeking an efficient procedure to find parameters that minimize the es mated trajectory errors. From the experimental results we draw conclusions as to both the efficiency of the op miza on methods, and the role of parcular parameters in the visual odometry system.

show abstract

Section: System Structurementioning

confidence: 99%

Section: System Structurementioning

confidence: 99%

Section: System Structurementioning

confidence: 99%

See 1 more Smart Citation

On the Efficiency of Population-Based Optimization in Finding Best Parameters for RGB-D Visual Odometry

Kostusiak¹,

Skrzypczyński²

2019

JAMRIS

Self Cite

View full text Add to dashboard Cite

show abstract

“…In comparison to SLAM, the VO is a simpler approach to localization based on the same data. In order to investigate and visualize how important are the drift reduction techniques for achieving the accurate trajectory estimation, a simple RGB-D VO pipeline has been proposed, based on the procedures available in the OpenCV library [16]. The RGB-D VO system searches salient point features in the RGB image from the current frame and tries to match them with the features from the previous frame.…”

Section: Rgb-d Vomentioning

confidence: 99%

“…Whereas the literature is rich in papers evaluating feature detectors and descriptors, also with respect to various aspects of mobile robot localization [7,16,28], very few authors studied the in luence of particular SLAM system architectures on the accuracy and reliability of robot trajectory estimation. Strasdat [34] compared several versions of his visual SLAM system, however, working mostly with the passive cameras.…”

Section: Introduc Onmentioning

confidence: 99%