We propose a video stabilization algorithm based on the rotation of a virtual sphere. Unlike traditional video stabilization algorithms relying on two-dimensional motion models or reconstruction of (3D) camera motions, the proposed virtual sphere model stabilizes video by projecting each frame onto the sphere and performing corrective rotations. Specifically, matching feature points between two adjacent frames are first projected onto two virtual spheres to obtain pairs of spherical points. Then, the rotation matrix between the previous and current frame is calculated. The resulting 3D rotation matrix sequence is used to represent the camera motion, and it is smoothed using the geodesic distance on a Riemannian manifold. Finally, the difference between the smoothed and original path allows obtaining the rotation matrix that causes camera jitter, and the virtual spheres are rotated reversely to suppress jitter. Experimental results show that the proposed algorithm can effectively reduce random jitter, outperforming state-of-the-art methods.
More and more electronic products use non-smooth surfaces to improve user experience. Computer vision based quality control is very important for these products. Since images taken from the non-smooth surfaces have complex micro textures and low contrasts, it is very challenging to detect defects in the images. To solve this problem, this paper proposes a training-free method which first enhances defects and then detect them accurately. At the phase of defect enhancement, pixel-level saliency is first calculated by two novel features named localglobal intensity difference and local intensity aggregation, and then an iterative enhancement approach named accumulated aggregation shifting (AAS) is proposed to shift each pixel’s intensity according to its saliency. At the phase of defect detection, two statistic models, including linear distribution or exponential distribution, are fitted by the shifting results of AAS at different iterations. Based on the fitted statistic models, defective pixels are defect-free pixels are accurately classified by risk minimization. Experimental results prove that the proposed approach is effective in detecting defects on non-smooth surfaces of real industrial products.
A method for detecting an object's motion in images that suffer from camera shake or images with camera egomotion
is proposed. This approach is based on edge orientation codes and on the entropy calculated from a histogram of the edge
orientation codes. Here, entropy is extended to spatio-temporal entropy. We consider that the spatio-temporal entropy
calculated from time-series orientation codes can represent motion complexity, e.g., the motion of a pedestrian. Our
method can reject false positives caused by camera shake or background motion. Before the motion filtering, object
candidates are detected by a frame-subtraction-based method. After the filtering, over-detected candidates are evaluated
using the spatio-temporal entropy, and false positives are then rejected by a threshold. This method could reject 79 to 96
[%] of all false positives in road roller and escalator scenes. The motion filtering decreased the detection rate somewhat
because of motion coherency or small apparent motion of a target. In such cases, we need to introduce a tracking method
such as Particle Filter or Mean Shift Tracker. The running speed of our method is 32 to 46 ms per frame with a 160×120
pixel image on an Intel Pentium 4 CPU at 2.8 GHz. We think that this is fast enough for real-time detection. In addition,
our method can be used as pre-processing for classifiers based on support vector machines or Boosting.
Image processing method that detects a particular moving object from an image by a fixed camera and tracking is
noticed in various fields and it is a very important subject. In this paper, we propose a moving object tracking method
that can cope with change of the area accompany the random walk movement of the moving object oneself and change
of the brightness arise from change of the environmental such as a masking or change of the illumination. Proposal
method can be robust processing for change of the illumination based on Orientation Code Matching that is
demonstrated that is robust for the masking or change of the illumination. And, using Motion Vector derived from a
continuity of the random walk model motion, under the condition that there are similar walk models, it can discriminate
the walk model and individually tracking. Through the some experiment, this paper inspects the effectiveness of our
proposed method.
This paper proposes a novel focus measure based on self-matching methods. A unique pencil-shaped profile is identified by comparing the similarity between patterns extracted around their neighborhood in each scene. Based on this, a new criterion function, CPV, is defined to evaluate focused or defocused scenes. OCM is recommended due to its invariance with regards to contrasts. Experiments using a telecentric lens are implemented to demonstrate the efficiency of proposed measure. Comparing OCM-based focus measure with conventional focus measures shows that OCM-based CPV is robust against illuminations. Using this method, pan-focused images are composed and depth information is represented.
This study aims to establish an error model of the stereo measurement system considering camera vibration.
At first, we verified the distribution of disparity error under the circumstance without the camera vibration and with the camera vibration. As the result, we found that we can approximate the distribution of disparity error by normal distribution under the circumstance without camera vibration and with camera vibration. And, the parameters of normal distribution are changed by the camera vibration.
The parameters of the distribution of the measurement error are average μ and standard deviation σ. The parameters of the camera vibration are considered amplitude A and frequency F. In order to verify relationships during the parameters of the distribution of measurement error and the parameters of the camera vibration, we experimented using the vibration testing system. We imposed simple harmonic motion to the stereo camera. In this paper, we use stereo camera Bumblebee. As the result of experiment, the camera vibration didn't affect average μ. We found positively correlation between standard deviation σ and amplitude A. And, we found negatively correlation between standard deviation σ and frequency F. We estimate the parameters of measurement error by the parameter of the camera vibration using these relationships. So, we establish the error model of the stereo measurement system. Moreover, we define existing probability of object using the parameter of measurement error.
Nighttime images of a scene from a surveillance camera have lower contrast and higher noise than their corresponding daytime images of the same scene due to low illumination. Denighting is an image enhancement method for improving nighttime images, so that they are closer to those that would have been taken during daytime. The method exploits the fact that background images of the same scene have been captured all day long with a much higher quality. We present several results of the enhancement of low quality nighttime images using denighting.
KEYWORDS: 3D metrology, Robots, Mobile robots, Fluctuations and noise, Stereoscopic cameras, Distance measurement, Statistical analysis, Information science, Agriculture, 3D image processing
Due to the area of the vineyard in Hokkaido is extremely large, it is very difficult and hard to eradicate weeds
by human being. In order to solve this problem, we developed a dynamic image measure technique, which can
be applied to the weeding robots in vineyards. The outstanding of this technique is that it can discriminate the
weed and the trunk correctly and efficiently. Meanwhile, we also attempt to measure the root of trunk accurately.
And a new method to measure the blocked trunk of grapes in vineyards has also been developed in this paper.
This paper has mainly discussed about two problems, object focusing and depth measurement. First, we propose a novel and robust scheme of image focusing by introducing a new measure of focusing based on Orientation code matching. A new evaluation function, named Complemental pencil volume, CPV, is defined and calculated to represent local sharpness of images, either in or out of focus, by comparing the similarity between any patterns extracted at the same position within their own scenes. An identified and unique maximum or peek, which of ill-condition scenes with low contrast observations. Experiments show that the OCM-based focusing is very robust to change in brightness, and to even more irregularities in the real imaging system, like dark condition. Second, based on this robust focusing technique, we applied it to an image sequence of an object surface to measure the depth of profile. A simple plane object surface has been implemented to demonstrate the basic approach. The results showed the successful and precision depth measurement of this object.
Quality of local regions in the scene could be deteriorated by the effects of ill-conditioned lighting effects and reflection. We develop a method to improve the quality of local regions. The local region of the image is firstly selected by an algorithm that is based on similarity evaluation of the region. Then the histogram
of brightness and saturation of the local region are expanded using the histogram equalization, both the
brightness and saturation are improved. In the next step, based on the distance from the point selected
by user, the improved data are combined with the original image. The local part of the image is naturally
merged with the surrounding scenes.
Recently, developing of image processing method which enables to track to moving objects on time series images
taken by a fixed camera is one of important subjects in the field of machine vision. Here, we try to consider
influences by change in brightness and change of region caused by moving objects, respectively. In this paper,
we introduce a new tracking method which can be reduced the influences by those changes. First, we use Radial
Reach Filter in order to detect the moving objects. In addition, the moving objects can be tracked by an image
processing based on information obtained by applying RRF and block division. Further, we propose a method in
the case that changes size of moving object by time progress. Finally, through experiments we show the validity
of our proposed method.
Instead of tachometer-type velocity sensors, an effective method of estimating real-time velocities based on a robust image matching algorithm is proposed to measure real-time velocities of agrimotors or working machines, such as sprayers and harvesters, driven by them in real farm fields.
It should be precise even they have any slipping, and stable and robust to many ill-conditions occurred in the real world farm fields, where the robust and fast algorithm for image matching, Orientation code matching is effectively utilized.
A prototype system has been designed for real-time estimation of velocities of the agrimotors and then the effectiveness has been verified through many frames obtained from the real fields in various weather and ground conditions.
This study aims to expand a measuring range of stereovision system. In the previous paper, the authors achieved a 3D motion capture system by using one camera with triangle markers, named as Mono-MoCap (MMC). MMC has two features. One is that MMC can measure 3D positions of subjects by using one camera on the basis of the perspective n point (PnP) problem. Another is that MMC does not need to recalibrate the camera parameters. In this paper, the authors apply MMC to binocular stereovision system for expanding its measurement range. MMC will solve the problem that stereovision system can not measure 3D positions of objects when even one camera does not capture the objects. In this study, 3D positions of three points where a geometrical relation each other is already-known will be measured by a stereovision. 3D position measurement by the stereovision is enabled by secondarily using MMC when a part of the object is hidden with one camera of the stereovision. Simulation and experimental results will show the effectiveness of the proposed method for expanding the measuring range of stereovision system.
We propose self-maintenance robot system as a method which realizes work for a long time without maintenance by the human workers. This system absorbs the change which occurs in robot's hardware by learning, and maintains working ability. We propose the two methods of learning changes in the physical information of the robot as methods which realizes the maintenance-free robot system. One is a method to learn robot's physical information based on the input and output information in the task practice from the no physical information of the robot by using a neural network which has a task common layer and a task independence layer. We use a neural network which has a task common layer and a task independence layer to learning. Other is a method to learn robot's physical information based on the difference in hoping action and actual action. In this report, we verify of these learning system by the computer simulation.
This paper introduces tracking control system with visual feedback to a moving object by using the measurement device which we developed. In order to recognize the moving object, we use two method, using cross-shape mark and Orientation Code Matching (OCM). And the measurement device is constructed PID control system with Extended Kalman Filter in order to track to object. Through the several experiments, we verify the percormance of recognition and tracking.
This paper aims to propose a fast image searching method from environmental observation images even in the presence of scale changes. A new scheme has been proposed for extracting feature areas as tags based on a robust image registration algorithm called Orientation code matching. Extracted tags are stored as template
images and utilized in tag searching. As the number of tags grows, the searching cost becomes a serious problem. Additionally, change in viewing positions cause scale change of an image and matching failure. In our scheme, richness in features is important for tag generation and the entropy is used to evaluate the diversity of edge
directions which are stable to scale change of the image. This characteristic contributes to limitation of searching area and reduction in calculation costs. Scaling factors are estimated by orientation code density which means the percentage of effective codes in fixed size tag areas. An estimated scaling factor is applied to matching a scale of template images to one of observation images. Some experiments are performed in order to compare computation time and verify effectiveness of estimated scaling factor using real scenes.
KEYWORDS: Detection and tracking algorithms, 3D acquisition, Head, Data modeling, 3D metrology, Distance measurement, Image processing, Data processing, Feature extraction, Error analysis
This paper presents a novel fast and high-accurate 3-D registration algorithm. The ICP (Iterative Closest Point) algorithm converges all the 3-D data points of two data sets to the best matching points with minimum evaluation values. This algorithm is broadly used, because it has good availability to many applications. But, it needs many computational costs and it is very sensible to error values. Because, it uses whole data points of two data sets and least mean square optimization. We had proposed the M-ICP algorithm, which is an extension of the ICP algorithm based on modified M-estimation for realization of robustness against outlying gross noise.
The proposed algorithm named HM-ICP (Hierarchical M-ICP) is an extension of the M-ICP with selecting region for matching and hierarchical searching of selected regions. In this algorithm, we select regions using evaluation of variance for distance values in the target region and homogeneous topological mapping.
Some fundamental experiments utilizing real data sets of 3-D measurement show effectiveness of the proposed method. We achieved more than 4-digits number reduction of computational costs and confirmed less than 0.1% error to the measurement distance.
A rotation-invariant template matching scheme using Orientation Code Difference Histogram (OCDH) is proposed. Orientation code features based on local distributions of pixel brightness are substantively robust against furious change in illumination plays a main role in designing the rotation-invariant matching algorithm. Since every difference between any pair of orientation codes is invariant in rotation of an image, we can elaborate a histogram feature by use of the differences, which can aggregate effective clues for searching rotated images through simple procedures. With gray scale images as targets, rotation angles of an image can be accurately estimated by the proposed method. It is fast and robust even in presence of some irregularities as brightness change by shading or highlighting. We propose a two-stage framework for realizing the rotation-invariant template matching based on OCDH. In the first stage, candidate positions are selected through evaluation of OCDH at every position, and then in the second stage, they are tested by use of a verification also based on orientation code features. The effectiveness of the proposed matching method has been shown through many kinds of experiments designed with real world images.
Feature extraction and tracking are widely applied in the industrial world of today. It is still an important topic in Machine Vision. In this paper, we present a new feature extraction and tracking method which is robust against illumination change such as shading and highlighting, scaling and rotation of objects. The method is composed mainly of two algorithms: Entropy Filter and Orientation Code Matching (OCM). The Entropy Filter points up areas of images being messy distribution of orientation codes. The orientation code is determined by detecting the orientation of maximum intensity change around neighboring 8 pixels. It is defined as simply integral values. We can extract good features to track from the images by using the Entropy Filter. And then, the OCM, a template matching method using the orientation code, is applied to track the features each frame. We can track the features robustly against the illumination change by using the OCM. Moreover, updating these features (templates) each frame allows complicated motions of tracked objects such as scaling, rotation and so on. In this paper, we report the details of our algorithms and the evaluations of comparison with other well-known feature extraction and tracking methods. As an application example, planer landmarks and face tracking is tried. The results of them are also reported in context.
In this paper, we propose a new method of object detection. In the past, there are various methods of object detection. Especially, the method of the background subtraction has the effectiveness. However, the methods based on brightness differences are easily influenced by change in lighting condition. In this paper, we use Radial Reach Filter (RRF). RRF is called as the effective method of the change in lighting conditions. However, RRF is not considered change that caused by moving objects on the background image. Then, we propose the new method of object detection that considered motion of the moving objects on the background image. And, we verify the effectiveness by the experiments using a time series image.
A novel successive learning algorithm is proposed for efficiently handling sequentially provided training data based on Test Feature Classifier (TFC), which is non-parametric and effective even for small data. We have proposed a novel classifier TFC utilizing prime test features (PTF) which is combination feature subsets for getting excellent performance. TFC has characteristics as follows: non-parametric learning, no mis-classification of training data. And then, in some real-world problems, the effectiveness of TFC is confirmed through way applications. However, TFC has a problem that it must be reconstructed even when any sub-set of data is changed.
In the successive learning, after recognition of a set of unknown objects, they are fed into the classifier in order to obtain a modified classifier. We propose an efficient algorithm for reconstruction of PTFs, which is formalized in cases of addition and deletion of training data. In the verification experiment, using the successive learning algorithm, we can save about 70% on the total computational cost in comparison with a batch learning. We applied the proposed successive TFC to dynamic recognition problems from which the characteristic of training data changes with progress of time, and examine the characteristic by the fundamental experiments. Support Vector Machine (SVM) which is well established in algorithm and on practical application, was compared with the proposed successive TFC. And successive TFC indicated high performance compared with SVM.
We propose an efficient template matching algorithm for binary image search. When we use template matching techniques, the computation cost depends on size of images. If we have large size images, we spend a lot of time for searching similar objects in scene image to template image. We design a scanning-type upper limit estimation that can be useful for neglect correlation calculation. For calculating the scanning-type upper limits, template and scene images are divided into two regions: R-region and P-region. In R-region, an upper limit of correlation coefficients can be derived as an interval estimation based on mathematical analysis of correlations of the object image and a pivot image. In P-region, another upper limit is formalized based on the number of white and black pixels in a template and the object image. By use of these upper limits, the scanning-type upper limit estimation of correlation coefficients can be formalized for the efficient matching algorithm. This upper limits estimation isn't over true values of correlation, so the accuracy of search by conventional search is the same as one by conventional search. The experiments with document images show the effectiveness and efficiency of the proposed matching algorithm. In these experiments, computation time by the proposed algorithm is between 5 and 20% compare of the conventional search.
We have proposed a brand-new noninvasive ultrasonic sensor for measuring muscle activities named as Ultrasonic Muscle Activity Sensor (UMS). In the previous paper, the authors achieved to accurately estimate joint torque by using UMS and electromyogram (EMG) which is one of the most popular sensors. This paper aims to realize to measure not only joint torque also joint angle by using UMS and EMG. In order to estimate torque and angle of a knee joint, muscle activities of quadriceps femoris and biceps femoris were measured by both UMS and EMG. These targeted muscles are related to contraction and extension of knee joint. Simultaneously, actual torque on the knee joint caused by these muscles was measured by using torque sensor. The knee joint angle was fixed by torque sensor in the experiment, therefore the measurement was in isometric state.
In the result, we found that the estimated torque and angle have high correlation coefficient to actual torque and angle. This means that the sensor can be used for angle estimation as well torque estimation. Therefore, it is shown that the combined use of UMS and EMG is effective to torque and angle estimation.
This study aims to realize a motion capture for measuring 3D human motions by using single camera. Although motion capture by using multiple cameras is widely used in sports field, medical field,
engineering field and so on, optical motion capture method with one camera is not established. In this paper, the authors achieved a 3D motion capture by using one camera, named as Mono-MoCap (MMC), on the basis of two calibration methods and triangle markers which each length of side is given. The camera calibration methods made 3D coordinates transformation parameter and a lens distortion parameter with Modified DLT method. The triangle markers enabled to calculate a coordinate value of a depth direction on a camera coordinate. Experiments of 3D position measurement by using the MMC on a measurement space of cubic 2 m on each side show an average error of measurement of a center of gravity of a triangle marker was less than 2 mm. As compared with conventional motion capture method by using multiple cameras, the MMC has enough accuracy for 3D measurement. Also, by putting a triangle marker on each human joint, the MMC was able to capture a walking motion, a standing-up motion and a bending and stretching motion. In addition, a method using a triangle marker together with conventional spherical markers was proposed. Finally, a method to estimate a position of a marker by measuring the velocity of the marker was proposed in order to improve the accuracy of MMC.
This paper aims to propose a new scheme for robust tagging for landmark definition in unknown circumstance using some qualitative evaluations based on Orientation Code representation and matching which has been proposed for robust image registration even in the presence of change in illumination and occlusion. Necessary characteristics for effective tags: richness, similarity, and uniqueness, are considered in order to design an algorithm for tag extraction. These qualitative considerations can be utilized to design simple and robust algorithm for tag definition in combination with the robust image registration algorithm.
KEYWORDS: Data modeling, Databases, Object recognition, Statistical analysis, Statistical modeling, Model-based design, Detection and tracking algorithms, Feature extraction, Process modeling, Monte Carlo methods
In order to gain generality, robustness and efficiency in search, a novel search is proposed based on a representation called `Depth aspect image' is proposed as a controllable two-dimensional representation of local depth distribution used in cooperation with a distinct `Voxel framing', which enables effective reference coordination without any prominent features, such as vertices or edges. A robust statistical estimator called `Least quantile of residuals' is furthermore introduced for robust matching, which can be utilized for both depth matching and model verification.
Since the proposed method is of model-based approach with possible views of local structures, the computation cost for matching has to be reduced by introducing random sampling and an effective hashing.
Experiments with real scenes show the effectiveness of the proposed method.
An efficient algorithm for searching similar images from databases of a large set of images is proposed. For designing the algorithm, all the correlation coefficient values of registered images are calculated in advance to online computation and they are memorized as a set of keys for efficient search. We theoretically derive an interval estimation of any correlation coefficient between an object image and arbitrary registered images in terms of a pivot image that is one of candidates of the unique solution image and can be simply selected. Using the interval estimations on all the other registered images, some conditions for selecting redundant images from the registered images can be derived and evaluated and then those who has any smaller similarity than the one computed by the pivot image can be skipped away from correlation computation, which efficiently enables to save a lot of computational cost for search. The algorithm was applied to real searching problems in an image database of 1,200 images taken from the real world and to search in multiple template matching problems, for example rotation invariant matching and a search problem on binary maps, resulting the efficiency of the proposed method for the real problems.
A framework for disigning robust registration algorithms is proposed. Robust image registration is demanded in environment of ill-conditions occured in the real world. Coding is a major mechanism for their purposes and the robust registration methods are designed through the following four phases: positional combination, code generation from brightness difference, definition of similarity or dissimilarity, and statistical modelling or analysis. Reformalization of increment sign correlation (ISC) and orientaion code matching (OCM) are described in this view point. And then the application of ISC to borehole measurement is presented as the one of the real world tasks.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.