Continuously Adaptive Mean Shift Research Articles

Assisted Operation of a Robotic Arm Based on Stereo Vision for Positioning near an Explosive Device

This document presents an assisted operation system of a robotic arm for positioning near an explosive device selected by the user through the visualization of the cameras on the screen. Two non-converging cameras mounted on the robotic arm in camera-in-hand configuration provide the three-dimensional (3D) coordinates of the object being tracked, using a 3D reconstruction technique with the help of the continuously adaptive mean shift (CAMSHIFT) algorithm for object tracking and feature matching. The inverse kinematics of the robot is implemented to locate the end effector close to the explosive so that the operator can perform the operation of grabbing the grenade more easily. Inverse kinematics is implemented in its geometric form, thus reducing the computational load. Tests conducted with various explosive devices verified the effectiveness of the system in locating the robotic arm in the desired position.

Research on Statistics of Passenger Transfer Flow at Urban Rail Transit Station

With the continuous advancement of urbanization, the urban rail transit network continues to be dense, and the number of transfer stations increases, resulting in an increase in the flow of interchange passengers. This paper introduces pedestrian detection into the statistics to realize real-time statistics of passenger transfer flow. This article first determines the research scenario and simplifies the transfer path. Then, the pedestrian detection model is constructed by using the histogram of gradient directions (HOG) and support vector machine (SVM), and combined with the continuous adaptive mean shift algorithm (CamShift) to design and realize the transfer path passenger flow statistics. Finally, the algorithm was trained and tested through the samples obtained at Xi’an Xiaozhai subway station, and the test results obtained have high accuracy.

Implementation of Camshift Target Tracking Algorithm Based on Hybrid Filtering and Multifeature Fusion

In recent years, the Mean shift algorithm has extensive applications in the field of video tracking. It has some advantages of low cost, small memory, and good tracking effect. However, there are some shortcomings in the existing algorithm; for example, it cannot produce adaptive changes as the target size changes. And when there are similar objects, it is prone to target positioning errors and tracking failures caused by occlusion. In this paper, an improved method of continuous adaptive change Mean shift (Camshift) for high-precision positioning and tracking is proposed. The traditional Camshift method only uses hue components in HSV to extract features. This paper uses the combination of H and S components in HSV space to build a two-dimensional color feature histogram and with the image’s LBP feature histogram to increase tracking accuracy. Meanwhile, for the sake of target occlusion and nonlinear changes in the tracking process, this paper introduces a Gaussian-Hermit particle filter that is updated by the Kalman filter. Experimental result demonstrates that the real-time performance of the proposal in this paper is better than Mean shift, Camshift, simple particle filter, and Kalman filter.

INDIVIDUAL TREE DETECTION FROM UAV LIDAR DATA IN A MIXED SPECIES WOODLAND

Abstract. The study evaluates five existing segmentation algorithms to determine the method most suitable for individual tree detection across a species-diverse forest: raster-based region growing, local maxima centroidal Voronoi tessellation, point-cloud level region growing, marker controlled watershed and continuously adaptive mean shift. Each of the methods has been tested twice over one mixed and five single species plots: with their parameters set as constant and with the parameters calibrated for every plot. Overall, continuous adaptive mean shift performs best across all the plots with average F-score of 0.9 with fine-tuned parameters and 0.802 with parameters held at constant. Raster-based algorithms tend to achieve higher scores in coniferous plots, due to the clearly discernible tops, which significantly aid the detection of local maxima. Their performance is also highly dependent on the moving size window used to detect the local maxima, which ideally should be readjusted for every plot. Crown overlap, suppressed and leaning trees are the most likely sources of error for all the algorithms tested.

Eye-to-hand robotic tracking and grabbing based on binocular vision

This paper aims to study eye-to-hand robotic tracking and grabbing based on binocular vision. Two cameras placed on different locations gave different three-dimensional coordinates of the object from a binocular vision. The robot then analyzed the dynamic vision acquired from the stereo cameras and tracked the object in three-dimensional space utilizing continuously adaptive mean shift (CAMSHIFT) algorithm. Afterward, inverse and forward kinematics were implemented to move the robotic arm to the appropriate position so that it can grab the object using the end effector. The inverse kinematics was analyzed by a geometric algorithm to reduce the computational burden. Consequently, the experimental results verified that the eye-to-hand system was able to track and grasp the target successfully.

The effect of illumination compensation methods with histogram back projection for camshift application

This paper presents the results of a factorial experiment performed to determine the effect of illumination compensation methods with histogram back projection to be used for object tracking algorithm continuous adaptive mean-shift (Camshift). Since Camshift tracking can be used for distance approximation of an object, a precise tracker algorithm is required. This study compared two types of illumination compensation methods using Design of Experiment (DOE) in the presence of illumination inconsistency. Based on the results, it was found that selecting two channels as reference in histogram back projection weakens the accuracy of Camshift tracking whereas the combination of both methods produces results that are more desirable. Keywords : object tracking; camshift; vision system; DOE.

Multi-technique object tracking approach- A reinforcement paradigm

In recent years, surveillance is fast becoming an essential part of everyday life in both commercial and residential environments due to the vital role it plays in the society. The existing tracking methods are limited by factors such as illumination variations, occlusions, camera movements, and background clutters among others. Hence, a robust multi-technique tracking method based on continuous adaptive mean shift (CamShift), corrected background weighted histogram, and unscented particle filter techniques is proposed. The mean square error statistic was used to evaluate the performance of the proposed method in terms of correctly estimating the path of a target object in video sequence acquired from three different scenarios. The obtained results showed that our proposed method has a significant improvement both quantitatively and qualitatively across all scenarios compared to the traditional CamShift and some existing methods. The proposed multi-technique approach might potentially improve the tracking capability of surveillance devices.

Vision Based Tracking of Moving Target in an Autonomous Ground Vehicle Framework

This paper proposes a vision-based algorithm to autonomously track a moving target with an autonomous ground vehicle. The proposed approach, which is to estimate the target’s position and orientation, is built on a robust colour based tracker using the Continuously Adaptive Mean Shift (CAMShift) algorithm. The object tracker can handle occlusions, lighting and environment effects in a single framework when combined with Multiple Kalman Filters. The information is then used from the visual tracker to control the position and yaw angle of the UAV in order to track the object and keep it in the field of view. The system is practically implemented and tested using the Arduino platform and a cheap low cost web based camera.

Real-Time Computer Vision System for Continuous Face Detection and Tracking

The ever-increasing number of traffic accidents due to a diminished driver’s vigilance level has become a problem of serious concern to society. With the ever growing traffic conditions, this problem will further deteriorate. For this issue, development of system which can actively monitors driver vigilance level and alert the driver for any insecure driving condition is essential. So this paper gives detailed information about driver vigilance level monitoring system. The ultimate goal of the system is to detect and alert the driver from insecure sleepy or low concentration driving condition. The system consists of two main modules including drivers face and eye detection module and drivers face tracking module. Viola Jones face detection with AdaBoost (AdaptiveBoosting) method and Circular Hough Transform technique are integrated in the drivers face and eye detection module. In the drivers face tracking module, CAMSHIFT (Continuously Adaptive Mean Shift) algorithm has been used for continuous face tracking of driver. The main components of the system consist of a video camera, a specially designed hardware system based on Raspberry Pi for real-time image processing and controlling the alarm system. In the proposed system, only one video camera is used in practice yet an achievement of fast and accurate detection results are obtained.

강인한 손가락 끝 추출과 확장된 CAMSHIFT 알고리즘을 이용한 자연스러운 Human-Robot Interaction을 위한 손동작 인식

In this paper, we propose a robust fingertip extraction and extended Continuously Adaptive Mean Shift (CAMSHIFT) based robust hand gesture recognition for natural human-like HRI (Human-Robot Interaction). Firstly, for efficient and rapid hand detection, the hand candidate regions are segmented by the combination with robust <TEX>$YC_bC_r$</TEX> skin color model and haar-like features based adaboost. Using the extracted hand candidate regions, we estimate the palm region and fingertip position from distance transformation based voting and geometrical feature of hands. From the hand orientation and palm center position, we find the optimal fingertip position and its orientation. Then using extended CAMSHIFT, we reliably track the 2D hand gesture trajectory with extracted fingertip. Finally, we applied the conditional density propagation (CONDENSATION) to recognize the pre-defined temporal motion trajectories. Experimental results show that the proposed algorithm not only rapidly extracts the hand region with accurately extracted fingertip and its angle but also robustly tracks the hand under different illumination, size and rotation conditions. Using these results, we successfully recognize the multiple hand gestures.

Hand Tracking and Hand Gesture Recognition for Human Computer Interaction

The aim of this paper is to present the methodology for hand tracking and hand gesture recognition. The detected hand and gesture can be used to implement the non-contact mouse. We had developed a MP3 player using this technology controlling the computer instead of mouse. In this algorithm, we first do a pre-processing to every frame which including lighting compensation and background filtration to reducing the adverse impact on correctness of hand tracking and hand gesture recognition. Secondly, YCbCr skin-color likelihood algorithm is used to detecting the hand area. Then, we used Continuously Adaptive Mean Shift (CAMSHIFT) algorithm to tracking hand. As the formula-based region of interest is square, the hand is closer to rectangular. We have improved the formula of the search window to get a much suitable search window for hand. And then, Support Vector Machines (SVM) algorithm is used for hand gesture recognition. For training the system, we collected 1500 hand gesture pictures of 5 hand gestures. Finally we have performed extensive experiment on a Windows XP system to evaluate the efficiency of the proposed scheme. The hand tracking correct rate is 96% and the hand gestures average correct rate is 95%.

A robust approach for multiple vehicles tracking using layered particle filter

Abstract Multiple vehicle targets tracking is one of the most challenging problems in Intelligent Transportation Systems. It is used for recognizing and understanding vehicle behaviors, especially suffering from illumination, scale, pose variations and occlusions. In this paper, we explore a robust tracking algorithm, combining deterministic and probabilistic methods, to solve this problem. We build a fusion observation model with color and local integral orientation descriptor, and give multiple vehicle targets model. In order to overcome the disadvantage of particle impoverishment, we propose a layered particle filter architecture embedding continuous adaptive mean shift, which considers both concentration and diversity of particles, and the particle set can better represent the posterior probability density. This paper also presents experiments using real video sequences to verify the proposed method.

Dynamical information fusion of heterogeneous sensors for 3D tracking using particle swarm optimization

This paper presents a new method for three dimensional object tracking by fusing information from stereo vision and stereo audio. From the audio data, directional information about an object is extracted by the Generalized Cross Correlation (GCC) and the object’s position in the video data is detected using the Continuously Adaptive Mean shift (CAMshift) method. The obtained localization estimates combined with confidence measurements are then fused to track an object utilizing Particle Swarm Optimization (PSO). In our approach the particles move in the 3D space and iteratively evaluate their current position with regard to the localization estimates of the audio and video module and their confidences, which facilitates the direct determination of the object’s three dimensional position. This technique has low computational complexity and its tracking performance is independent of any kind of model, statistics, or assumptions, contrary to classical methods. The introduction of confidence measurements further increases the robustness and reliability of the entire tracking system and allows an adaptive and dynamical information fusion of heterogenous sensor information.

Automatic CamShift tracking algorithm based on multi-feature

Since the Continuously Adaptive MeanShift(CamShift)tracking algorithm only adopts color as feature,which would result in tracking error,an improved algorithm based on feature fusion was proposed.The presented algorithm detected the moving target automatically using an improved background subtraction method.Object model combined color and gradients orientation features,and weighted the reliability of features.It also overcame the possible CamShift invalidity in the situation of some similar color objects.The experimental results show that the algorithm enhances the reliability and robustness of tracking.

Autonomous Hovering and Landing of a Quad-rotor Micro Aerial Vehicle by Means of on Ground Stereo Vision System

On ground stereo vision system is used for autonomous hovering and landing of a quadrotor Micro Aerial Vehicle (MAV). This kind of system has an advantage to support embedded vision system for autonomous hovering and landing, since an embedded vision system occasionally gives inaccurate distance calculation due to either vibration problem or unknown geometry of the landing target. Color based object tracking by using Continuously Adaptive Mean Shift (CAMSHIFT) algorithm was examined. Nonlinear model of quad-rotor MAV and a PID controller were used for autonomous hovering and landing. The result shows that the Camshift based object tracking algorithm has good performance. Additionally, the comparison between the stereo vision system based and GPS based autonomous hovering of a quad-rotor MAV shows that stereo vision system has better performance. The accuracy of the stereo vision system is about 1 meter in the longitudinal and lateral direction when the quad-rotor flies in 6 meters of altitude. In the same experimental condition, the GPS based system accuracy is about 3 meters. Additionally, experiment on autonomous landing gives a reliable result.

Multiobject tracking by isMHI combining with CamShift

To realize the real-time tracking of multiple objects in video scene,a tracking method based on increased-step Motion History Image(isMHI)and Continuously Adaptive Mean Shift(CAMSHIFT) is proposed.The moving objects entering the video scene could be tracked based on the isMHI,and they could be locked and monitored based on CAMSHIFT instead once they turn to stay,moreover the joint algorithm between the moving state and resting state was also defined.Experimental results show that this method combines with their advantages and overcomes the faults from isMHI based and CAMSHIFT,respectively.So not only the continual,stable and real-time tracking of moving objects can be realized,but also resting objects in video scene with the error rate remains low.Thus,it can be applied to monitoring the interesting objects in the security or the traffic.

A real-time fuzzy reasoning based control system for catching a moving goldfish

Designing a real-time visual tracking system to catch a goldfish is a complex task because of the large amount of streaming video data that must be transmitted and processed immediately when tracking the goldfish. Usually, building such visual servoing systems requires the application of high-cost specialized hardware and the development of complicated visual control software. In this paper, a novel low-cost, real-time visual servo control system is presented. The system uses stereo vision consisting of two calibrated cameras to acquire images of the goldfish, and applies the continuously adaptive mean shift (CAMSHIFT) vision-tracking algorithm to provide feedback of a fish’s real-time position at a high frame rate. It then employs a 5-axis robot manipulator controlled by a fuzzy reasoning system to catch the fish. This visual tracking and servoing system is less sensitive to lighting influences and thus performs more efficiently. Experiments with the proposed method yielded very good results, as the system’s real-time 3D vision successfully tracked two fish and guided the manipulator, which has a net attached to its end effector, to catch one of them.

Machine vision-based real-time early flame and smoke detection

This paper proposes a novel real-time machine video-based flame and smoke detection method that can be incorporated with a surveillance system for early alerts. Automatic monitoring systems use the motion history detection algorithm to register the possible flame and smoke position in a video and then analyze the spectral, spatial and temporal characteristics of the flame and smoke regions in the image sequences. The spectral probability density is represented by comparing the flame and smoke color histogram model, where HSI color spaces are used. The spatial probability density is represented by computing the flame and smoke turbulent phenomena with the relation of perimeter and area. Statistical distribution of the spectral and spatial probability density is weighted with the fuzzy reasoning system to give the potential flame and smoke candidate region. The temporal probability density is represented by extracting the flickering area with level crossing and separating the alias objects from the flame and smoke region. Then, the continuously adaptive mean shift (CAMSHIFT) vision tracking algorithm is employed to provide feedback of the flame and smoke real-time position at a high frame rate. Experimental results under a variety of conditions show that the proposed method is capable of detecting flame and smoke reliably.

A Real-time Visual Servo Architecture for a Golden Fish Catching System

Designing a real-time visual tracking system to catch golden fish is a complex task due to difficulties involved in the precise determination of a fish's relative 3D positions in the water tank. In this paper we present a new visual servo control system using stereo vision to acquire the 3D positions of the golden fish. The continuously adaptive mean shift (CAMSHIFT) algorithm and optical flow vision tracking methods are applied to provide, at a high frame-rate, the real-time feedback of the position and motion trajectory of the fish being tracked. A visual servo architecture for controlling two low-cost, well calibrated cameras for tracking the golden fish and a 5-axis robot manipulator for catching a specified golden fish are constructed using standard off-the-shelf hardware and open source software. The designed visual tracking and servo system is less sensitive to lighting influences and thus performs more efficiently. Experiments of the proposed method yielded very good results as the real-time 3D vision was able to successfully track two fish and guide the manipulator to catch one of them with a net attached to the manipulator's end-effector.