Multiple-object Scenes Research Articles

Non-textured objects visual SLAM using pola-rization 3D reconstruction

The common visual SLAM technology is to obtain the scene image information through monocular and binocular cameras, and then process the camera sensor data to get the restored map. However, these camera sensors are not sensitive enough to non-textured scenes to restore the map. It is easy to cause large deviation in image acquisition and cannot restore the corresponding shape. Therefore, in order to address these challenges, this paper proposes a SLAM method that uses polarization camera to obtain polarization characteristics of the target and then calculate the depth information to realize image reconstruction. Specifically, in the SLAM process, we use a monocular polarization camera to acquire an image sequence, and the phase angle and normal vector of each pixel obtained by polarization are used to recover the depth of the textured object, then multi-view normal vector constraints are fused to carry out 3D reconstruction, and finally, in the process of depth propagation, polarized light field information is used to constrain the propagation of non-textured regions. We respectively verified the indoor and outdoor shooting data of single and multiple object scenes, and the reconstruction results are compared with the existing DSO, ORB-SLAM and other algorithms, which shows that the effect of our visual SLAM method is better than that of the existing common SLAM method in reconstructing the surface shape of non-textured objects.

300 GHz radar object recognition based on deep neural networks and transfer learning

For high-resolution scene mapping and object recognition, optical technologies such as cameras and LiDAR are the sensors of choice. However, for future vehicle autonomy and driver assistance in adverse weather conditions, improvements in automotive radar technology and the development of algorithms and machine learning for robust mapping and recognition are essential. In this study, the authors describe a methodology based on deep neural networks to recognise objects in 300 GHz radar images using the returned power data only, investigating robustness to changes in range, orientation and different receivers in a laboratory environment. As the training data is limited, they have also investigated the effects of transfer learning. As a necessary first step before road trials, they have also considered detection and classification in multiple object scenes.

3D modeling of multiple-object scenes from sets of images

This paper proposes a new approach for multi-object 3D scene modeling. Scenes with multiple objects are characterized by object occlusions under several views, complex illumination conditions due to multiple reflections and shadows, as well as a variety of object shapes and surface properties. These factors raise huge challenges when attempting to model real 3D multi-object scene by using existing approaches which are designed mainly for single object modeling. The proposed method relies on the initialization provided by a rough 3D model of the scene estimated from the given set of multi-view images. The contributions described in this paper consists of two new methods for identifying and correcting errors in the reconstructed 3D scene. The first approach corrects the location of 3D patches from the scene after detecting the disparity between pairs of their projections into images. The second approach is called shape-from-contours and identifies discrepancies between projections of 3D objects and their corresponding contours, segmented from images. Both unsupervised and supervised segmentations are used to define the contours of objects.

Modelling Visual Search with the Selective Attention for Identification Model (VS-SAIM): A Novel Explanation for Visual Search Asymmetries

In earlier work, we developed the Selective Attention for Identification Model (SAIM [16]). SAIM models the human ability to perform translation-invariant object identification in multiple object scenes. SAIM suggests that central for this ability is an interaction between parallel competitive processes in a selection stage and a object identification stage. In this paper, we applied the model to visual search experiments involving simple lines and letters. We presented successful simulation results for asymmetric and symmetric searches and for the influence of background line orientations. Search asymmetry refers to changes in search performance when the roles of target item and non-target item (distractor) are swapped. In line with other models of visual search, the results suggest that a large part of the empirical evidence can be explained by competitive processes in the brain, which are modulated by the similarity between target and distractor. The simulations also suggest that another important factor is the feature properties of distractors. Finally, the simulations indicate that search asymmetries can be the outcome of interactions between top-down (knowledge about search items) and bottom-up (feature of search items) processing. This interaction in VS-SAIM is dominated by a novel mechanism, the knowledge-based on-centre-off-surround receptive field. This receptive field is reminiscent of the classical receptive fields but the exact shape is modulated by both, top-down and bottom-up processes. The paper discusses supporting evidence for the existence of this novel concept.

An iterative multiresolution scheme for SFM with missing data: Single and multiple object scenes

Most of the techniques proposed for tackling the Structure from Motion problem (SFM) cannot deal with high percentages of missing data in the matrix of trajectories. Furthermore, an additional problem should be faced up when working with multiple object scenes: the rank of the matrix of trajectories should be estimated. This paper presents an iterative multiresolution scheme for SFM with missing data to be used in both the single and multiple object cases. The proposed scheme aims at recovering missing entries in the original input matrix. The objective is to improve the results by applying a factorization technique to the partially or totally filled in matrix instead of to the original input one. Experimental results obtained with synthetic and real data sequences, containing single and multiple objects, are presented to show the viability of the proposed approach.

The distribution of attention within objects in multiple-object scenes: Prioritization by spatial probabilities and a center bias

This study investigates how the attentional distribution within objects is affected by spatial probabilities, bias toward objects' centers (Alvarez & Scholl, 2005), and object motion. In a multiple-object tracking task, observers tracked line objects while simultaneously detecting probes appearing on the objects. Experiments 1 and 2 manipulated the probabilities of probes appearing at the centers and ends of objects. Overall, probe detection was better at centers than at ends, but it was also affected by probe location probabilities; when probe locations were 100% certain, the center advantage was eliminated. Experiment 3 manipulated rotational, translational, and size-change components of object motion. The center advantage still occurred with stationary objects, and its magnitude was not affected by different motion types. These results indicate that attention is biased toward the centers of objects in multiple-object scenes, both for stationary and moving objects. They also imply that attentional prioritizations based on spatial probabilities can accompany moving objects.

Describing Visual Scenes Using Transformed Objects and Parts

We develop hierarchical, probabilistic models for objects, the parts composing them, and the visual scenes surrounding them. Our approach couples topic models originally developed for text analysis with spatial transformations, and thus consistently accounts for geometric constraints. By building integrated scene models, we may discover contextual relationships, and better exploit partially labeled training images. We first consider images of isolated objects, and show that sharing parts among object categories improves detection accuracy when learning from few examples. Turning to multiple object scenes, we propose nonparametric models which use Dirichlet processes to automatically learn the number of parts underlying each object category, and objects composing each scene. The resulting transformed Dirichlet process (TDP) leads to Monte Carlo algorithms which simultaneously segment and recognize objects in street and office scenes.

Binocular information in the control of prehensile movements in multiple-object scenes.

Recent evidence suggests that the visual control of prehension may be less dependent on binocular information than has previously been thought. Studies investigating this question, however, have generally only examined reaches to single objects presented in isolation, even though natural prehensile movements are typically directed at objects in cluttered scenes which contain many objects. The present study was designed, therefore, to assess the contribution of binocular information to the control of prehensile movements in multiple-object scenes. Subjects reached for and grasped objects presented either in isolation or in the presence of one, two or four additional 'flanking' objects, under binocular and monocular viewing conditions. So that the role of binocular information could be clearly determined, subjects made reaches both in the absence of a visible scene around the target objects (self-illuminated objects presented in the dark) and under normal ambient lighting conditions. Analysis of kinematic parameters indicated that the removal of binocular information did not significantly affect many of the major indices of the transport component, including peak wrist velocity. However, peak grip apertures increased and subjects spent more time in the final slow phase of movement, prior to grasping the object, during monocularly guided reaches. The dissociation between effects of binocular versus monocular viewing on transport and grasp parameters was observed irrespective of the presence of flanking objects. These results therefore further question the view that binocular vision is pre-eminent in the control of natural prehensile movements.

Robust topology-adaptive snakes for image segmentation

‘Snakes’-based segmentation techniques have a variety of applications in computer vision. Traditional snakes however are well known to be topologically inflexible. They are incapable of dealing with more complicated object shapes as well as multiple-object scenes since the snakes require that the topology of object structures of interest must be known in advance. This paper introduces a robust topology-adaptive snake, based on the attractable snake model [6] , to extend the snakes' topological adaptability. Three embedded schemes: the robust self-looping process, the efficient contour-merging and the improved adaptive interpolation scheme, are involved. Experiment results show that the new snake model is able to; consistently evolve towards its target objects, handle topological changes (i.e. splitting or merging) automatically as necessary and conform to more complicated geometries and topologies, without restrictive requirements on the initial conditions of the snake or on its deformation movement.

Effects of hippocampal and parietal cortex lesions on the processing of multiple-object scenes.

Effects of hippocampal and parietal cortex lesions on the processing of multiple-object scenes.

Effects of hippocampal and parietal cortex lesions on the processing of multiple-object scenes.

Rats were taught 1 of 3 types of multiple-object scene discrimination. Manipulations included either replacing (object based), displacing (space based), or replacing and displacing (object/space based) 1 object in the scene. Once trained, rats received hippocampal, parietal cortex, or cortical control lesions and then were retested. Parietal cortex- and hippocampal-lesioned rats displayed deficits on both the space and object/space tasks but not the object task. The hippocampal-lesioned group's deficit on the object/space task was only transient, whereas the impairments of the parietal cortex-lesioned rats were stable for both tasks. The parietal cortex-lesioned rats sustained difficulty with the object/space task may indicate an inability to switch cognitive strategies.

Recovering the 3D structure of tubular objects from stereo silhouettes

Although silhouette-based image understanding is attractive from an engineering viewpoint, recovering 3D shape from a single stereo pair of silhouette images of a generic multiple-object scene is a highly underconstrained problem. With respect to a gray-level-based approach, the ambiguities in stereo matching and the loss of data due to mutual visual occlusions are even more severe. These problems are alleviated when the observed objects can be assumed to belong to some restricted class. In this paper we consider the case of almost vertical tubular objects (AVTOs), i.e. generalized cylinders with some restrictions on their axis' shape and pose relative to the stereo pair. This restriction, together with the assumption that the scene must be explained with the minimum number of objects consistent with the observations, allows one to devise an effective reconstruction algorithm. The object shape/location parameters are estimated by recursive least-squares (Kalman) filtering. Constrained blind tracking is performed on the occluded sections by feeding the filters with the most likely parameter values compatible with the constraints induced by the observed images. The case of AVTOs with circular cross-section is analyzed in some detail, with examples taken from an actual implementation of the algorithm in the field of agricultural automation.

Point Signatures: A New Representation for 3D Object Recognition

Few systems capable of recognizing complex objects with free-form (sculptured) surfaces have been developed. The apparent lack of success is mainly due to the lack of a competent modelling scheme for representing such complex objects. In this paper, a new form of point representation for describing 3D free-form surfaces is proposed. This representation, which we call the point signature, serves to describe the structural neighbourhood of a point in a more complete manner than just using the 3D coordinates of the point. Being invariant to rotation and translation, the point signature can be used directly to hypothesize the correspondence to model points with similar signatures. Recognition is achieved by matching the signatures of data points representing the sensed surface to the signatures of data points representing the model surface. The use of point signatures is not restricted to the recognition of a single-object scene to a small library of models. Instead, it can be extended naturally to the recognition of scenes containing multiple partially-overlapping objects (which may also be juxtaposed with each other) against a large model library. No preliminary phase of segmenting the scene into the component objects is required. In searching for the appropriate candidate model, recognition need not proceed in a linear order which can become prohibitive for a large model library. For a given scene, signatures are extracted at arbitrarily spaced seed points. Each of these signatures is used to vote for models that contain points having similar signatures. Inappropriate models with low votes can be rejected while the remaining candidate models are ordered according to the votes they received. In this way, efficient verification of the hypothesized candidates can proceed by testing the most likely model first. Experiments using real data obtained from a range finder have shown fast recognition from a library of fifteen models whose complexities vary from that of simple piecewise quadric shapes to complicated face masks. Results from the recognition of both single-object and multiple-object scenes are presented.

A SURFACE FEATURE ATTRIBUTED HYPERGRAPH REPRESENTATION FOR 3-D OBJECT RECOGNITION

A surface feature hypergraph (SFAHG) representation is proposed for the recognition and localization of three-dimensional objects. The hypergraph representation is shown to be viewpoint independent thus resulting in substantial savings in terms of memory for the object model database. The resulting hypergraph matching algorithm integrates both, relational and the rigid pose constraint in a consistent unified manner. The matching algorithm is also shown to have a polynomial order of complexity even in multiple-object scenes with instances of objects partially occluding each other. An algorithm for incrementally constructing the hypergraph representation of an object model from range images of the object taken from different viewpoints is also presented. The hypergraph matching and the hypergraph construction algorithms are shown to be capable of correcting errors in the initial segmentation of the range image. The hypergraph construction algorithm and the matching algorithm are tested on range images of scenes containing multiple three-dimensional objects with partial occlusion.

Circuits of the mind

1: The Approach. 2: Biological Constraints. 2.1: Introduction. 2.2: The Neocortex. 2.3: Pyramidal Neurons. 3: Computational Laws. 3.1: Introduction. 3.2: Three Sources of Complexity. 4: Cognitive Functions. 4.1: Introduction. 4.2: Boolean Functions. 4.3: Learning. 4.4: The Nature of Concepts. 4.5: Experimental Psychology. 5: The Neuroidal Model. 5.1: Programmable Models. 5.2: Neuroids. 5.3: Timing. 6: Knowledge Representations. 6.1: Positive Knowledge Representations. 6.2: Vicinal Algorithms. 6.3: Frontier Properties and Storing New Items. 6.4: Frontier Properties and Associations. 6.5: Hashing. 7: Unsupervised Memorization. 7.1: An Algorithm. 8: Superivsed Memorization. 8.1: Introduction. 8.2: A Simple Algorithm. 8.3: A Second Algorithm. 9: Supervised Inductive Learning. 9.1: Introduction. 9.2: Pac Learning. 9.3: Learning Conjunctions. 9.4: Learning Disjunctions. 9.5: Learning Linear Threshold Functions. 10: Correlational Learning. 10.1: An Algorithm. 10.2: Computing with Numerical Values. 11: Objects and Relational Expressions. 11.1: Multiple Object Scenes. 11.2: Relations. 11.3: Timed Conjunctions. 11.4: Memorizing Expressions Containing Relations. 11.5: Memorizing New Relations. 11.6: Discussion. 12: Systems Questions. 12.1: Introduction. 12.2: General Organizational Principles. 12.3: Compatibility of Mechanisms. 13: Reasoning. 13.1: Introduction. 13.2: Relfex Reasoning. 13.3: Simple Reflex Reasoning. 13.4: Compound Reflex Reasoning. 13.5: Nonmonotonic Phenomena. 14: More Detailed Neural Models. 14.1: Implementing Vicinal Algorithms. 14.2: A Laminar Model. 14.3: A Columnar Model. 14.4: Sparser Random Graphs. 14.5: Another Columnar Model. 15: Afterword. Notes. Exercises. References. Index of Notation. Index

Parallelizing object recognition on the hypercube

Abstract In this paper a parallel Interpretation Tree search algorithm for object recognition on the Intel iPSC/2 hypercube multicomputer is presented. The input to the algorithm is a set of sensory measurements consisting of sensor locations and the directions of the surface normal. Such data is typical of sparse range or tactile sensors. At this time the algorithm is capable of handling two-dimensional objects or three-dimensional objects located on a plane with all sensory measurements parallel to the plane. The objects are typically those which a robot would encounter in an industrial scene during the process of automated inspection or automated assembly. The algorithm is able to handle polyhedra or objects which can be approximated as piecewise combination of polyhedra. Both single- and multiple-object scenes are considered. Three strategies for mapping the Interpretation Tree search process on the hypercube are considered. The performance of the algorithm for each of these mapping strategies is examined in terms of parameters such as speed-up, processor utilization, inter-processor communication overhead and load sharing between processors. It is shown that the requirement for uniform load sharing leads to increased inter-processor communication. Experimental results are presented and future directions are outlined.