Reliable Object Detection Research Articles

Detection and Recognition of Traffic Planar Objects Using Colorized Laser Scan and Perspective Distortion Rectification

Reliable detection and recognition of planar objects including traffic sign, street sign, and road surface in dynamic cluttered natural scenes are a big challenge for self-driving cars. In this paper, we propose a comprehensive method for planar object detection and recognition. First, the data association of LIDAR and camera is set up to acquire colorized laser scans, which simultaneously contain both color and geometrical information. Second, we combine three color spaces of RGB, HSV, and CIE L*a*b* with laser reflectivity as an aggregation-based feature vector. Third, the 3-D geometrical characteristics of planar objects that contain planarity, size, and aspect ratio are exploited to further reduce false alarm. Fourth, in order to increase robustness to any viewpoint variation, we present a new virtual camera-based rectification method to synthesize fronto-parallel views of refined object descriptors in 3-D space. Finally, experimental results achieved under a variety of challenging conditions show that integration of color space aggregation and laser reflectivity is superior to individuals. Specifically, the proposed perspective distortion rectification method remarkably eliminates false recognition error by 45.5%. Overall, the detection rate of our comprehensive method has up to 95.87% and the recognition rate even reaches 95.07% for traffic signs ranging within 100 m, with about 33.25 ms average running time per frame.

Towards reliable object detection in noisy images

Image noise is a common problem frequently caused by insufficient lighting, low-quality cameras, image compression and other factors. While low image quality is expected to degrade results of visual recognition, most of the current methods and benchmarks for object recognition, such as Pascal Visual Object Classes Challenge and Microsoft Common Objects in Context Challenge, focus on relatively high-quality images. Meanwhile, object recognition in noisy images is a common problem in surveillance and other domains. In this work we address object detection in noisy images and propose a novel low-cost method for image denoising. When combined with the standard Deformable Parts Model and Regions with Convolutional Neural Network object detectors, our method shows improvements of object detection under varying levels of image noise. We present a comprehensive experimental evaluation and compare our method to other denoising techniques as well as to standard detectors re-trained on noisy images. Results are presented for the common Pascal Visual Object Classes benchmark for object detection and KAIST Multispectral Pedestrian Detection Benchmark with the real noise presence in night images.

Object Detection and Recognition for Assistive Robots: Experimentation and Implementation

Technological advances are being made to assist humans in performing ordinary tasks in everyday settings. A key issue is the interaction with objects of varying size, shape, and degree of mobility. Autonomous assistive robots must be provided with the ability to process visual data in real time so that they can react adequately for quickly adapting to changes in the environment. Reliable object detection and recognition is usually a necessary early step to achieve this goal. In spite of significant research achievements, this issue still remains a challenge when real-life scenarios are considered. In this article, we present a vision system for assistive robots that is able to detect and recognize objects from a visual input in ordinary environments in real time. The system computes color, motion, and shape cues, combining them in a probabilistic manner to accurately achieve object detection and recognition, taking some inspiration from vision science. In addition, with the purpose of processing the input visual data in real time, a graphical processing unit (GPU) has been employed. The presented approach has been implemented and evaluated on a humanoid robot torso located at realistic scenarios. For further experimental validation, a public image repository for object recognition has been used, allowing a quantitative comparison with respect to other state-of-the-art techniques when realworld scenes are considered. Finally, a temporal analysis of the performance is provided with respect to image resolution and the number of target objects in the scene.

A Combined Voxel and Particle Filter-Based Approach for Fast Obstacle Detection and Tracking in Automotive Applications

In this paper, a new method for real-time detection, motion estimation, and tracking of generic obstacles using just a 3-D point cloud and odometry information as input is presented. In this approach, a simplification of the world is done using voxels, supported by a particle filter-based 3-D object segmentation and a motion estimation scheme. That combination of techniques leverages a fast and reliable object detection, providing also motion speed and direction information. Four detailed studies have been performed in order to assess the suitability of the method, two of them related to the parameterization of the method and its input point cloud. Another one compares the tracking and detection results with other state-of-the-art methods. Last tests are intended for the characterization of the execution times required. Results are encouraging, with a high detection rate, low error rate, and real-time capable computing performance. In the attached video, it is possible to observe the behavior of the method, both using a stereovision and a light-detection and ranging generated point clouds as an input.

Multi-Section Sensing and Vibrotactile Perception for Walking Guide of Visually Impaired Person.

Electronic Travel Aids (ETAs) improve the mobility of visually-impaired persons, but it is not easy to develop an ETA satisfying all the factors needed for reliable object detection, effective notification, and actual usability. In this study, the authors developed an easy-to-use ETA having the function of reliable object detection and its successful feedback to the user by tactile stimulation. Seven ultrasonic sensors facing in different directions detect obstacles in the walking path, while vibrators in the tactile display stimulate the hand according to the distribution of obstacles. The detection of ground drop-offs activates the electromagnetic brakes linked to the rear wheels. To verify the feasibility of the developed ETA in the outdoor environment, walking tests by blind participants were performed, and the evaluation of safety to ground drop-offs was carried out. From the experiment, the feasibility of the developed ETA was shown to be sufficient if the sensor ranges for hanging obstacle detection is improved and learning time is provided for the ETA. Finally, the light-weight and low cost ETA designed and assembled based on the evaluation of the developed ETA is introduced to show the improvement of portability and usability, and is compared with the previously developed ETAs.

Touchscreen Surface Based on Interaction of Ultrasonic Guided Waves With a Contact Impedance

In this paper, a touchscreen device is proposed, based on guided-wave reflection and transmission induced by the presence of an object. The principle uses the advantages of other acoustic wave devices in terms of simplicity and applicability to any thin surface but is not subject to classical drawbacks (single touch, sensitivity to scratches or contaminant, and impossibility to follow motion of contact point). The theoretical interaction of guided waves with contact impedance is first derived in order to define the requirements of the sensor in terms of frequency range, mode, sensor type and location, and embedded electronics. Design criteria and experimental validation on a small prototype ( $200 \times 250$ mm) are proposed to demonstrate the potential of the approach for simple, robust, and reliable contact detection of pointlike or extended objects for consumer electronics or biomedical applications.

Lean histogram of oriented gradients features for effective eye detection

Reliable object detection is very important in computer vision and robotics applications. The histogram of oriented gradients (HOG) is established as one of the most popular hand-crafted features, which along with support vector machine (SVM) classification provides excellent performance for object recognition. We investigate dimensionality deduction on HOG features in combination with SVM classifiers to obtain efficient feature representation and improved classification performance. In addition to lean HOG features, we explore descriptors resulting from dimensionality reduction on histograms of binary descriptors. We consider three-dimensionality reduction techniques: standard principal component analysis, random projections, a computationally efficient linear mapping that is data independent, and locality preserving projections (LPP), which learns the manifold structure of the data. Our methods focus on the application of eye detection and were tested on an eye database created using the BioID and FERET face databases. Our results indicate that manifold learning is beneficial to classification utilizing HOG features. To demonstrate the broader usefulness of lean HOG features for object class recognition, we evaluated our system’s classification performance on the CalTech-101 dataset with favorable outcomes.

対象物と類似物の識別に有効な画素群を用いたテンプレートマッチング

This paper introduces a fast and reliable object detection method for situations in which a target object is surrounded by similar objects. Sample image sets of the target object and similar objects are prepared in advance. And pixels that maximize the separability between these two classes are automatically selected. Robust image matching for similar objects is achieved using these pixels. In addition, processing time for image matching is reduced by using a small number of pixels. Through experiments using 400 actual images, a recognition success rate of 98.5% was achieved by using approximately 0.5% of whole pixels of the template image.

Recursive On-Line <formula formulatype="inline"> <tex Notation="TeX">${(2{\rm D})}^2{\rm PCA}$</tex></formula> and Its Application to Long-Term Background Subtraction

In this paper, we propose a novel background subtraction method which enables reliable detection of foreground objects in a long surveillance video stream. Recently, although much progress has been made in the field of background subtraction, there are still challenging scenarios (e.g., high frequency motion of dynamic texture, non-stationary motion of camera, abrupt changes of illumination, etc.) in the long surveillance videos in which even state-of-the-art methods are often prone to fail. To cope with these challenging scenarios effectively, in the proposed method, a background model is initialized in a low-dimensional subspace and then updated periodically based on a novel recursive on-line ${(2{\rm D})}^2{\rm PCA}$ algorithm developed in this paper. Moreover, a threshold map is also updated in a scene-adaptive manner for labeling each pixel in a scene either foreground or background independently. Based on this on-line framework, the background of a surveillance video stream is reconstructed over time, thereby facilitating the detection of foreground objects reliably. In extensive experiments, we demonstrate that the proposed background subtraction method can cope with the aforementioned challenging scenarios more favorably than the state-of-the-art methods.

Towards Scene Understanding with Detailed 3D Object Representations

Current approaches to semantic image and scene understanding typically employ rather simple object representations such as 2D or 3D bounding boxes. While such coarse models are robust and allow for reliable object detection, they discard much of the information about objects’ 3D shape and pose, and thus do not lend themselves well to higher-level reasoning. Here, we propose to base scene understanding on a high-resolution object representation. An object class—in our case cars—is modeled as a deformable 3D wireframe, which enables fine-grained modeling at the level of individual vertices and faces. We augment that model to explicitly include vertex-level occlusion, and embed all instances in a common coordinate frame, in order to infer and exploit object-object interactions. Specifically, from a single view we jointly estimate the shapes and poses of multiple objects in a common 3D frame. A ground plane in that frame is estimated by consensus among different objects, which significantly stabilizes monocular 3D pose estimation. The fine-grained model, in conjunction with the explicit 3D scene model, further allows one to infer part-level occlusions between the modeled objects, as well as occlusions by other, unmodeled scene elements. To demonstrate the benefits of such detailed object class models in the context of scene understanding we systematically evaluate our approach on the challenging KITTI street scene dataset. The experiments show that the model’s ability to utilize image evidence at the level of individual parts improves monocular 3D pose estimation w.r.t. both location and (continuous) viewpoint.

Detection and Tracking of Dynamic Objects by Using a Multirobot System: Application to Critical Infrastructures Surveillance

The detection and tracking of mobile objects (DATMO) is progressively gaining importance for security and surveillance applications. This article proposes a set of new algorithms and procedures for detecting and tracking mobile objects by robots that work collaboratively as part of a multirobot system. These surveillance algorithms are conceived of to work with data provided by long distance range sensors and are intended for highly reliable object detection in wide outdoor environments. Contrary to most common approaches, in which detection and tracking are done by an integrated procedure, the approach proposed here relies on a modular structure, in which detection and tracking are carried out independently, and the latter might accept input data from different detection algorithms. Two movement detection algorithms have been developed for the detection of dynamic objects by using both static and/or mobile robots. The solution to the overall problem is based on the use of a Kalman filter to predict the next state of each tracked object. Additionally, new tracking algorithms capable of combining dynamic objects lists coming from either one or various sources complete the solution. The complementary performance of the separated modular structure for detection and identification is evaluated and, finally, a selection of test examples discussed.

Fusion of cooperative localization data with dynamic object information using data communication for preventative vehicle safety applications

Abstract. Cooperative sensors allow for reliable detection, classification and localization of objects in the vehicle's surroundings – even without a line-of-sight contact to the object. The sensor principle is based on a communication signal between the vehicle and a transponder attached to the object of interest – a pedestrian, for example. Thereby, localization information is gathered by measuring the round-trip time-of-flight (RTOF) and evaluating the angle-of-arrival (AOA) of the incident signal. After that, tracking algorithms are used to recover the kinematic state of the object providing a basis for situation assessment. This paper investigates possibilities and benefits of extending this principle by the communication of information from inertial sensors which are locally attached to the transponder. Furthermore, this paper presents a robust approach for fusing the localization data with dynamic object information using the Dempster-Shafer theory. The approach is evaluated by performing real-world experiments for the analysis of pedestrian accidents.

Vision-Based Haptic Feedback for Remote Micromanipulation in–SEM Environment

This article presents an intuitive environment for remote micromanipulation composed of both haptic feedback and virtual reconstruction of the scene. To enable nonexpert users to perform complex teleoperated micromanipulation tasks, it is of utmost importance to provide them with information about the 3-D relative positions of the objects and the tools. Haptic feedback is an intuitive way to transmit such information. Since position sensors are not available at this scale, visual feedback is used to derive information about the scene. In this work, three different techniques are implemented, evaluated, and compared to derive the object positions from scanning electron microscope images. The modified correlation matching with generated template algorithm is accurate and provides reliable detection of objects. To track the tool, a marker-based approach is chosen since fast detection is required for stable haptic feedback. Information derived from these algorithms is used to propose an intuitive remote manipulation system that enables users situated in geographically distant sites to benefit from specific equipments, such as SEMs. Stability of the haptic feedback is ensured by the minimization of the delays, the computational efficiency of vision algorithms, and the proper tuning of the haptic coupling. Virtual guides are proposed to avoid any involuntary collisions between the tool and the objects. This approach is validated by a teleoperation involving melamine microspheres with a diameter of less than 2 μ m between Paris, France and Oldenburg, Germany.

Motion Detection: A Biomimetic Vision Sensor Versus a CCD Camera Sensor

We compare a traditional CCD imaging sensor with a biomimetic vision sensor (based on Musca domestica, the common house fly) with regard to detecting small moving objects at various speeds and levels of contrast. The patent-pending biomimetic vision sensor described is shown to be capable of reliable object detection in spite of challenging movement and low contrast conditions in which a typical commercial CCD camera sensor failed to properly detect the object.

Multi-Purpose Environment Awareness Approach for Single Line Laser Scanner in a Small Rotorcraft UA

The work presents an environment awareness approach for a small rotorcraft unmanned aircraft (UA) which operates at low height using a single line laser scanner which enables a height estimation with a concurrent detection of ground fixed obstacles. The approach is suitable for small UA which are not able to carry complex and heavy 3D laser scanner mountings having additional drives or mirrors. It works without using external reference systems like DGPS. The approach was especially developed for a mission of the "International Micro Air Vehicle Conference" outdoor contest, where it is the aim to fly through a 6x6m artificial gate. The sensor data processing enables the height estimation above ground as well as the detection of obstacles in order to meet the mission's goal. The height estimation enables a near-ground flight to prevent a collision with a top boundary of the gate, and a terrain following. The obstacle detection senses the pillars of the gate and finds a safe way through the narrow gate passage. The development and optimisation of the mounting and the sensor processing, as well as the validation, was realized under operational conditions with manual remote control (RC) helicopter flights and virtual flights at a simulation environment. The results of the experiments show that with this approach the mission can be fulfilled as a reliable ground estimation and object detection is ensured.

Template Match Object Detection for Inertial Navigation Systems

This paper devoted to propose template match object detection for inertial navigation systems (INS). The proposed method is an image processing technique to improve the precision of the INS for detecting and tracking the ground objects from flying vehicles. Template matching is one of the methods used for ground object detection and tracking. Robust and reliable object detection is a critical step of object recognition. This paper presents a proposed mathematical morphological template matching method for detection and tracking of ground objects. Our focus is on flying systems equipped with camera to capture photos for the ground and recognize it. The proposed method is independent on the altitude or the orientation of the object. The algorithm is simulated using Matlab program and the numerical experiments are shown which verify the object detection for a wide range altitude and orientation. The results show superiority of this method for identifying and recognizing the ground objects.

Predicting Millimeter Wave Radar Spectra for Autonomous Navigation

Millimeter Wave (MMW) radars are currently used as range measuring devices in applications such as automotive driving aids (Langer and Jochem, 1996), (Rohling and Mende, 1996), the mapping of mines (Brooker et al., 2005) and autonomous field robotics (Brooker, 2001), (Langer, 1996). This recent interest is largely due to the advantages MMW radars offer over other range measuring sensors, as their performance is less affected by dust, fog, rain or snow and ambient lighting conditions. MMW radars can provide received signal strength values, at all discrete range intervals, within the working range of the radar (Clark and Durrant-Whyte, 1997), (Scheding et al., 2002). The received power versus range spectra hence contain useful range to target information, but are also corrupted by noise. User defined stochastic algorithms can then be implemented, which exploit this rich data to improve object detection and mapping performance. This is in contrast to many other range measuring devices which typically internally threshold received signals, to provide single hard decisions only, on the estimated range to objects (Mullane, 2007). This paper addresses the issues of predicting the power-range spectra from MMW radars which use the Frequency Modulated Continuous Wave (FMCW) range estimation technique. This is important for two reasons. First, in automotive and autonomous robotic applications, such sensors are used in conjunction with vehicle navigation and map state estimation filters. This is so that (typically uncertain) vehicle motion knowledge can be optimally fused with the noisy sensor information, to infer estimates of the state of interest (typically, the vehicle's pose (position and orientation) and/or information of the surrounding object locations). Hence, it is essential that predicted power versus range spectra can be computed, to apply a Bayesian recursive estimation framework, based on previous measurements, and uncertain vehicle motion information. Second, it is extremely useful to be able to simulate MMW radar data, given certain environmental configurations. This aids the development of reliable object detection algorithms, based on theoretical sensor and noise models, which can then be applied more effectively to real MMW radar data.

A General Framework for Multi-Human Tracking

The task of reliable detection and tracking of multiple objects becomes highly complex for crowded scenarios. In this paper, a robust framework is presented for multi-Human tracking. The key contribution of the work is to use fast calculation for mean shift algorithm to perform tracking for the cases when Kalman filter fails due to measurement error. Local density maxima in the difference image - usually representing moving objects - are outlined by a fast non-parametric mean shift clustering procedure. The proposed approach has the robust ability to track moving objects, both separately and in groups, in consecutive frames under some kinds of difficulties such as rapid appearance changes caused by image noise and occlusion.

A General Framework for Multi-Human Tracking using Kalman Filter and Fast Mean Shift Algorithms

The task of reliable detection and tracking of multiple objects becomes highly complex for crowded scenarios. In this paper, a robust framework is presented for multi-Human tracking. The key contribution of the work is to use fast calculation for mean shift algorithm to perform tracking for the cases when Kalman filter fails due to measurement error. Local density maxima in the difference image - usually representing moving objects - are outlined by a fast non-parametric mean shift clustering procedure. The proposed approach has the robust ability to track moving objects, both separately and in groups, in consecutive frames under some kinds of difficulties such as rapid appearance changes caused by image noise and occlusion.

Efficient and Low-Complexity Surveillance Video Compression Using Backward-Channel Aware Wyner-Ziv Video Coding

Video surveillance has been widely used in recent years to enhance public safety and privacy protection. A video surveillance system that deals with content analysis and activity monitoring needs efficient transmission and storage of the surveillance video data. Video compression techniques can be used to achieve this goal by reducing the size of the video with no or small quality loss. State-of-the-art video compression methods such as H.264/AVC often lead to high computational complexity at the encoder, which is generally implemented in a video camera in a surveillance system. This can significantly increase the cost of a surveillance system, especially when a mass deployment of end cameras is needed. In this paper, we discuss the specific considerations for surveillance video compression. We present a surveillance video compression system with low-complexity encoder based on Wyner-Ziv coding principles to address the tradeoff between computational complexity and coding efficiency. In addition, we propose a backward-channel aware Wyner-Ziv (BCAWZ) video coding approach to improve the coding efficiency while maintaining low complexity at the encoder. The experimental results show that for surveillance video contents, BCAWZ can achieve significantly higher coding efficiency than H.264/AVC intra coding as well as existing Wyner-Ziv video coding methods and is close to H.264/AVC inter coding, while maintaining similar coding complexity with intra coding. This shows that the low motion characteristics of many surveillance video contents and the low-complexity encoding requirement make our scheme a particularly suitable candidate for surveillance video compression. We further propose an error resilience scheme for BCAWZ to address the concern of reliable transmission in the backward-channel, which is essential to the quality of video data for real-time and reliable object detection and event analysis.