Efficient 3D Objects Recognition Using Multifoveated Point Clouds

Abstract

Technological innovations in the hardware of RGB-D sensors have allowed the acquisition of 3D point clouds in real time. Consequently, various applications have arisen related to the 3D world, which are receiving increasing attention from researchers. Nevertheless, one of the main problems that remains is the demand for computationally intensive processing that required optimized approaches to deal with 3D vision modeling, especially when it is necessary to perform tasks in real time. A previously proposed multi-resolution 3D model known as foveated point clouds can be a possible solution to this problem. Nevertheless, this is a model limited to a single foveated structure with context dependent mobility. In this work, we propose a new solution for data reduction and feature detection using multifoveation in the point cloud. Nonetheless, the application of several foveated structures results in a considerable increase of processing since there are intersections between regions of distinct structures, which are processed multiple times. Towards solving this problem, the current proposal brings an approach that avoids the processing of redundant regions, which results in even more reduced processing time. Such approach can be used to identify objects in 3D point clouds, one of the key tasks for real-time applications as robotics vision, with efficient synchronization allowing the validation of the model and verification of its applicability in the context of computer vision. Experimental results demonstrate a performance gain of at least 27.21% in processing time while retaining the main features of the original, and maintaining the recognition quality rate in comparison with state-of-the-art 3D object recognition methods.

High-Level Path Planning for an Autonomous Sailboat Robot Using Q-Learning

“…Other strategies can also be devised for the sailboat robot to follow the generated path, with dynamica recalculation of the trajectory, if necessary, from each point. Approaches including visual attention features [39,40] and natural or artificial marks could also be devised for following a planned trajectory. Currently, these strategies are being implemented to work under ROS (Robotic Operating System) in order for them to be tested in the real N-Boat or sailboat robot ( Figure 17) during field trials.…”

Dynamic multifoveated structure for real-time vision tasks in robotic systems

“…A multifoveated (3D) approach inspired by models of (2D) multifoveation [1,4,5,27,47] and using the MMF [14] has been recently developed for dealing with point clouds to reduce the processing time in the detection of objects [15,16]. While this approach avoids the processing of redundant regions, it differs from our proposal, because it removes the intersection between levels of the same fovea structure, not between multiple foveae, as this is necessary in point cloud processing.…”

“…Among those works, there are some that perform the technique in real time, but using dedicated hardware (FPGA) to allow parallel (array) processing. Notice that this replication causes an intersection between the fovea levels if it is not previously treated and the only works in the literature dealing with this redundancy of information are the ones of Camails et al [1,27], Rodríguez et al [47], and Oliveira et al [15,16]. Oliveira's work deals with multifoveation in point clouds using a completely different approach for that, based on boxes.…”

“…Reducing the amount of visual data while keeping essential information can be done using the technique known as foveation [1][2][3][4][5][6][7][8][9][10][11][12][13]. This technique provides reduction of 2D [8,12] or 3D [14][15][16] data for facilitating further computations necessary for extraction of features, thus allowing the execution of visual tasks in realtime. Also known as multiresolution foveation, it is basically achieved by applying an image transformation from the spatial domain to obtain a dry structure in the multiresolution feature domain [3,8,14].…”

“…In such application, the foveation method should comply with a pleasant appearance for human visualization purposes, after reconstruction. Other applications, which are the main motivation of our research, are real-time processing for robotics vision [7][8][9][10][11]24] and real-time objects recognition [13][14][15][16]25]. In such applications, there are limited resources available to process all tasks at a given real-time slot, mainly considering a general purpose robot or 3D camera.…”