Rakhi Mutha

Rakhi Mutha

Auto-generated profile, based on publicly available information

Rakhi Mutha

Amity University, India

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 November 2023
Verified
Conference Paper
Open access
UMUdit Mamodiya
Udit Mamodiya
RMRakhi Mutha
Rakhi Mutha
Harish Reddy GantlaHarish Reddy Gantla
Harish Reddy Gantla
SPSunil Kr Pandey
Sunil Kr Pandey
SASayed Sayeed Ahmad
Sayed Sayeed Ahmad
ABAbhishek Bhattacherjee
Abhishek Bhattacherjee
Data mining uses artificial intelligence and other methods to extract hidden potential information from large data sets, which provides an effective way to obtain valuable knowledge from a large amount of information. Data mining is also ubiquitous in the process of using deep learning to solve point cloud registration tasks. Global feature extraction and rigid body transformation estimation are two critical stages of uncorrelated point cloud registration. Mining the rich information hidden in the two stages is one of the essential tasks of point cloud registration. However, the recently proposed methods tend to ignore low-dimensional local features when extracting global features, resulting in the loss of a large amount of point cloud information, which makes the accuracy of solving transformation parameters in the subsequent rigid body transformation estimation stage unsatisfactory. In this paper, a feature mining network based on multi-dimensional information fusion is proposed, which thoroughly mines the high-dimensional global and low-dimensional local information in the point cloud, and effectively makes up for the lack of local features in the worldwide feature extraction stage of point cloud registration. Dual quaternions are used to estimate pose in the rigid body transformation estimation stage, which can simultaneously represent rotation and translation in a standard frame, providing a compact and accurate representation for pose estimation. Finally, extensive experiments on the ModelNet40 dataset show that compared with the existing state-of-the-art point cloud registration methods without corresponding points, the proposed method can achieve higher accuracy and is more robust to noise.