担当区分：筆頭著者,　責任著者   掲載種別：研究論文（国際会議プロシーディングス）  

researchmap

掲載種別：研究論文（国際会議プロシーディングス）  

researchmap

記述言語：日本語   掲載種別：研究論文（学術雑誌）   出版者・発行元：一般社団法人 電気学会  

<p>The price of crops depends on the grade. The grade is determined from various factors such as color, shape, size, and scratches. It takes a lot of time and effort if farmers decide the grade. This study aims to automate shape grade judgement. In the prior method, binarization and contour extraction are performed on the tomato image to define features called flatness, variant, sharpness, and irregularity, and a threshold is set for each grade. However, due to the correlation between the features, the grade judgement has not been made with sufficient accuracy. In this study, we propose several methods focusing on contour features and compare these accuracies.</p>

DOI： 10.1541/ieejeiss.140.1134

CiNii Books

CiNii Research

researchmap

担当区分：筆頭著者,　責任著者   掲載種別：研究論文（学術雑誌）  

© 2019, The Author(s). Identification and quality assurance of stem cells cultured in heterogeneous cell populations are indispensable for successful stem cell therapy. Here we present an image-processing pipeline for automated identification and quality assessment of human keratinocyte stem cells. When cultivated under appropriate conditions, human epidermal keratinocyte stem cells give rise to colonies and exhibit higher locomotive capacity as well as significant proliferative potential. Image processing and kernel density estimation were used to automatically extract the area of keratinocyte colonies from phase-contrast images of cultures containing feeder cells. The DeepFlow algorithm was then used to calculate locomotion speed of the colony area by analyzing serial images. This image-processing pipeline successfully identified keratinocyte stem cell colonies by measuring cell locomotion speed, and also assessed the effect of oligotrophic culture conditions and chemical inhibitors on keratinocyte behavior. Therefore, this study provides automated procedures for image-based quality control of stem cell cultures and high-throughput screening of small molecules targeting stem cells.

DOI： 10.1038/s41598-019-55279-4

Scopus

PubMed

researchmap

記述言語：英語   掲載種別：研究論文（学術雑誌）   出版者・発行元：John Wiley and Sons Inc.  

This paper reports the regression analysis of the horizontal motion of a car driver's foot from a brake pedal to an accelerator pedal using machine learning methods, namely, neural network (NN), radial basis function network (RBFN), random forest (RF), and support vector regression (SVR) models, in order to predict the timing of when the driver steps on the accelerator from an web camera movie. The approximation performances of these models are compared based on root mean squared error (RMSE) and the estimated timing when the driver's foot reaches the accelerator pedal. From the results, the RF model could approximate the driver's foot motion most precisely. © 2018 Institute of Electrical Engineers of Japan. Published by John Wiley &amp
Sons, Inc.

DOI： 10.1002/tee.22614

Scopus

J-GLOBAL

researchmap

記述言語：日本語   掲載種別：研究論文（研究会，シンポジウム資料等）  

researchmap

記述言語：日本語   出版者・発行元：FIT(電子情報通信学会・情報処理学会)運営委員会  

CiNii Books

researchmap

記述言語：英語   掲載種別：研究論文（学術雑誌）   出版者・発行元：WILEY-BLACKWELL  

In image analysis, it is important to detect lines (straight lines) that represent the edges of an object in the image. The most popular technique for line detection is the Hough transform. In this transform, peaks resulting in the accumulator array that are obtained through a voting procedure in the parameter space represent strong evidence that a corresponding line exists in the image. In the voting procedure, large numbers of votes that are far from the peaks are unnecessary and reduce the speed of the transform. In this paper, we propose a method to speed up this process by introducing a procedure in image space that removes lines except for limited directions. This procedure makes it possible to reduce the voting range and to remove unnecessary votes. Experimental results confirm that the proposed method performs effectively.

DOI： 10.1002/ecj.11677

Web of Science

Scopus

researchmap

記述言語：英語   掲載種別：研究論文（国際会議プロシーディングス）   出版者・発行元：IEEE  

This paper describes estimation of the inverse model by multi-layered neural network. The back-propagation rule requires a sensitivity function of a system. If the system has uncertainly, then we can not calculate the sensitivity function. Hence, we propose a learning rule based on particle swarm optimization (PSO) combining with simultaneous perturbation. PSO and simultaneous perturbation are suitable for estimation of the inverse model with uncertainly, because they can update by only value of the objective function. PSO has a capability of finding a global minimum and simultaneous perturbation can search local area efficiently. We introduce two adaptation method of the combination ratio. One of them is to adapt it depending on the distance from gbest. The other is to adapt it depending on the value of the objective function. The proposed method are investigated using inverse kinematics problem. The simulation results show that the proposed methods obtain the more accurate inverse model.

DOI： 10.1109/SOCPAR.2015.7492782

Web of Science

researchmap

記述言語：英語   掲載種別：研究論文（国際会議プロシーディングス）   出版者・発行元：IEEE  

We consider estimation of an inverse model for the uncertain systems. Multi-layered neural network (NN) can approximate any continuous nonlinear mapping, then we use NN in order to represent the inverse model. The the back-propagation rule, which is commonly used for updating the weights, requires the sensitivity function of the system. However, we can not calculate this function because of uncertainty of the systems. Hence, we apply particle swarm optimization (PSO) to update the weights. PSO is suitable for learning NN, because it dose not require the derivative of the objective function. This paper introduces a novel parameter automation strategy in order to overcome the premature convergence. The acceleration coefficients are adaptively varies with the distance between the particle and the gbest which is the best solution ever found by the swarm. We can maintain the diversity of the swarm and search a better solution around the gbest efficiently. We compare the proposed method with the basic PSO and TVAC PSO, which is proposed by Ratnaweera, through the simulation of the inverse kinematics problem. The proposed method can obtain the accurate inverse model than the basic PSO and TVAC PSO.

DOI： 10.1109/SICE.2014.6935194

Web of Science

Scopus

researchmap

記述言語：日本語   掲載種別：研究論文（学術雑誌）   出版者・発行元：The Institute of Electrical Engineers of Japan  

In image analysis, it is important to detect lines (straight lines) which represent edges of an object in the image. The most popular technique to detect lines is Hough Transform. In the transform, resulting peaks in the accumulator array which are gotten by a voting procedure in the parameter space represent strong evidence that a corresponding line exists in the image. In the voting procedure, a large number of votes which are far from the peaks are unnecessary and it makes the transform slow. In this paper, we propose a speed-up method by introducing a procedure in image space which removes lines except for the limited directional lines. Owing to the procedure, we can reduce voting range and remove the unnecessary vote. From the experimental results, we confirm that the proposed method has good performance.

DOI： 10.1541/ieejeiss.133.1539

Scopus

researchmap

掲載種別：研究論文（国際会議プロシーディングス）   出版者・発行元：Springer  

This paper investigates the use of particle swarm optimization (PSO) with repeating splits and merges at predetermined intervals. After a split, there is no exchange of information between the particles, which belong to different swarms. Hence, even though one particle may be trapped in a local minimum, the others are not affected. If the other particles find a better solution, the trapped particle can escape the local minimum when the particles are merged. In order to verify the efficacy of the proposed method, we applied it to the learning of a neural network for solving the inverse kinematics problem of a manipulator with uncertain parameters. The back-propagation rule requires the Jacobian of the forward kinematics, but this cannot be calculated due to uncertainties. Because PSO does not require the derivative of the objective function, it is suitable for this problem. A simulation result shows that the proposed method can obtain more accurate inverse kinematics than either global best (gbest) PSO or local best (lbest) PSO. © 2013 Springer-Verlag.

DOI： 10.1007/978-3-642-38658-9_10

Scopus

researchmap

掲載種別：研究論文（学術雑誌）  

Recently, virtual reality (VR) environments where virtual objects can be accurately operated and the real world is closely mimicked have been constructed using haptic devices and physics simulators. To control virtual objects in physics simulators by operating haptic devices, it is necessary to connect the haptic pointer serving as the human hand in the VR environment and the grip of the device actually operated by users. As a means of realizing this, virtual coupling, which connects the two using a spring-damper model, can be used. In virtual coupling, devices may oscillate and their fidelity may decrease depending on the coupling parameters; therefore, application developers have empirically determined the coupling parameters for each application. In this study, we propose methods of determining the optimal coupling parameters by genetic algorithm.

Scopus

researchmap

記述言語：日本語   出版者・発行元：画像電子学会  

DOI： 10.11371/iieej.41.528

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：一般社団法人 画像電子学会  

本稿では，正の一次元フロー数の投票と，その結果に基づいた多方向への計算軸の設置を繰り返すことで，移動物体を追跡する手法を提案する．この手法では，一次元フローを用いて不良設定問題を回避し，多方向に計算軸を設置して追跡可能な動きを拡張する．更に，急激な方向変化に対応する機構も導入する．実画像を用いた追跡実験を行い，一次元フローによる移動物体の追跡が可能で あること，急激な方向変化への対応も可能であることを確認した．

DOI： 10.11371/iieej.37.720

researchmap

記述言語：英語   掲載種別：研究論文（国際会議プロシーディングス）   出版者・発行元：IEEE  

Although optical flow is useful for tracking of a moving object, estimation by the gradient method is ill-posed problem. In order to avoid ill-posedness, we have proposed a tracking method via one dimensional optical flow (ID flow), which is calculated on the straight line (calculation axis). The calculation axes are spanned several direction based on the estimated motion. This method votes with the number of positive ID flow and updates the calculation axes depending on the voting result. We update these axes every five frames to achieve stable estimation of motion. Hence, it is impossible to track a object which moves fast because this object moves to the outside of the region of. interest. If the calculation axes are updated every frame to track a object which moves fast, then this method tends to estimate the wrong direction due to vibration of the camera or partial occlusion.
We propose a tracking method via ID flow using queue. We prepare queues of which the length is five for each direction. The number of positive flow is enqueued. Direction which has maximum value is regarded as the moving direction. We can achieve stable tracking even if a object moves fast, because the calculation axes are updated every frame and estimation of the moving direction is based on the information of the past five frames. In experiment for real image sequences, the proposed method can track a object which moves fast.

DOI： 10.1109/ICARCV.2008.4795896

Web of Science

researchmap

記述言語：英語   掲載種別：研究論文（国際会議プロシーディングス）   出版者・発行元：IEEE  

We consider solving the inverse kinematics problem by forward-propagation rule with high order term. The goal signal is derived by the Newton-like method and the correction of weights is calculated by ridge regression. Hence, the learning times would be reduced if we can realize the goal signal accurately because this signal is derived by Newton-like method. We propose adjusting the regularization parameter of ridge regression depending on the high order term. The experimental result shows decrease of the learning times without loss of the accuracy of the inverse kinematics model.

Web of Science

researchmap

記述言語：日本語   出版者・発行元：一般社団法人電子情報通信学会  

画像から直線を抽出する有効な手法としてハフ変換が知られているが,ハフ平面(ρ-θパラメータ空間)の離散化による抽出精度の問題がある.また,直線を構成する点群の配置が非常に疎でばらつきがある場合,ハフ変換の性質上うまく直線を抽出できないという問題もある.このようなハフ変換の苦手とする直線を適切に抽出する手法としてLMedSハフ変換が提案されている.LMedSハフ変換は直線の評価値としてLMedS値を用いており,ハフ変換の多数決原理とは異なったアルゴリズムにより直線を適切に近似して抽出することが可能である.ところが,LMedSハフ変換ではLMedS値の計算量が多いため,その処理時間に問題がある.そこで本論文では,LMedS値の計算量を削減することで高速に直線を抽出する手法を提案し,その有効性を実験により示す.

researchmap

記述言語：英語   掲載種別：研究論文（学術雑誌）  

The Hough transform is known as a technique for detecting image components (primitives) such as lines or circles. We previously proposed a "primitive detection method using local polar coordinates," which can quickly detect primitives by reducing the dimensionality of the vote array of this Hough transform. However, this method has the property that the primitive detection speed is affected by the position and quantity of noise that is contained in the image (this property is called "noise dependency"), and this property had become a problem when the method was actually used. Therefore, in this paper, we propose a solution for this noise dependency problem. The proposed method sets a window in the vicinity of a local pole origin and determines whether or not a local pole can be set according to an operation (multiple votes per pixel) within this window. From various types of experimental results, we show that the proposed method is highly effective even though it is simple. © 2007 Wiley Periodicals, Inc.

DOI： 10.1002/scj.20359

Scopus

researchmap

researchmap

記述言語：英語   掲載種別：研究論文（国際会議プロシーディングス）   出版者・発行元：IEEE  

The optical flow is a useful tool for the tracking of a moving object. Estimation of the optical How based on the gradient method is an ill-posed problem. In order to avoid this ill-posed problem, we proposed a tracking method using a one-dimensional optical flow, which is calculated on a straight line (called the calculation axis) spanning several directions. However, the motion of the observer was not considered. In this paper, we propose object tracking by a one-dimensional optical flow under a rotating observer. The apparent motion of a stationary environment object should be eliminated for calculating the one-dimensional optical flow. Hence, we introduce the detection method of a moving object by mapping, which converts the motion of a stationary environment object into a linear signal trajectory. We calculate the one-dimensional optical flow by using pixels, which belong to the moving object, to eliminate the apparent motion of the stationary environment object. In order to verify the efficacy of the proposed method, simulation is performed using synthesized images. The proposed method successfully tracks the moving object when the observer rotates at a constant angular velocity.

DOI： 10.1109/ICARCV.2006.345264

Web of Science

researchmap

記述言語：英語   掲載種別：研究論文（国際会議プロシーディングス）   出版者・発行元：IEEE  

Estimation of an inverse kinematics model is important in robot manipulator control. Multilayered neural networks that can express a nonlinear mapping are applied in order to acquire the inverse kinematics model. Several estimation methods using neural networks have been proposed, and we herein consider the forward-propagation (FP) rule. This method is based on the following steps. First, the goal signal, which corresponds to the supervisor signal at the hidden layer and the output layer, is derived by the Newton-like method. Second, the updating of weights is realized by linear multiple regression. The regression coefficients express the adding correction of the weights. In the Fl? rule, it is important to realize the goal signal exactly. However, it is difficult to realize this requirement because the correction of the goal signal in the hidden layer, which corresponds to the objective variable, is expressed by the linear combination of the desired trajectory, which corresponds to the describing variable, because the neurons in the input layer are decided by the number of control outputs.
In this paper, we propose the FP rule with a high-order term. The layer, which is constructed with the high-order term of the input to the neural network, is inserted between the input layer and the hidden layer. In addition, we derive the goal signal in this scheme. In order to verify the efficacy of the FP rule with a high-order term, we apply this rule to the inverse kinematics problem for the two-link manipulator, which moves on the horizontal plane. The simulation results show that the proposed method obtains a more accurate inverse kinematics model, compared to the FP rule without the high-order term.

DOI： 10.1109/ICARCV.2006.345375

Web of Science

researchmap

記述言語：日本語   出版者・発行元：一般社団法人電子情報通信学会  

直線や円などの画像構成要素(プリミティブ)を抽出する手法としてハフ変換が知られている.筆者らは先に,このハフ変換の投票配列の次元数を少なくすることで高速にプリミティブを抽出できる「ローカル極座標を用いたプリミティブの抽出法」を提案した.しかしながらこの方法には,「プリミティブの抽出速度が,画像に含まれる雑音の位置と量に影響を受ける」という性質(この性質を「雑音依存性」と呼ぶ)があり,これが実際にこの方法を利用する際の問題となっていた.そこで本論文では,この雑音依存性に対する解決策を提案する.提案する方法は,ローカル極の原点付近にウィンドウを設定し,このウィンドウ内の演算(1点多投票)でローカル極設定の可否を判断するというもので,簡便な方法であるにもかかわらず高い効果を発揮することを種々の実験結果から明らかにしている.

CiNii Books

researchmap

researchmap

出版者・発行元：The Society of Instrument and Control Engineers  

Iterative learning control (ILC) obtains a desired input that exactly generates a desired output through repetitions of the similar tasks. The adjoint-type ILC, which is based on the gradient method, can obtain the bounded desired input for a finite dimensional and non-minimum phase system. This aspect suggests that the adjoint-type ILC can achieve good output tracking for a flexible arm, which is a infinite dimensional and non-minimum phase system. In this paper, we propose an update law and give a convergence condition for the adjoint-type ILC applied to a single-link flexible arm. In order to verify the effectiveness of the adjoint-type ILC for the single-link flexible arm, an experiment is carried out. Several researchers have proposed a iterative learning controller for the flexible arm by using assumed modes method. In this paper, a simple model of the single-link flexible arm is used because ILC dose not require the accurate model. Experiments result show that, even if the single-link flexible arm is modeled by the simple method and has some uncertainties, good output tracking is achieved. Moreover, pre-actuation, which is remarkable aspect of the adjoint-type ILC, is observed.

DOI： 10.9746/sicetr1965.39.35

researchmap

researchmap

researchmap

記述言語：日本語   出版者・発行元：計測自動制御学会  

DOI： 10.9746/sicetr1965.36.1117

CiNii Books

researchmap

researchmap

出版者・発行元：The Society of Instrument and Control Engineers  

Iterative learning control (ILC) obtains a desired input trajectory by repeating trials. Stable Inversion constructs a bounded input for desired output trajectory by using the non-causal inverses for non-minimum phase systems. In this paper, we investigate a relationship between the ILC using adjoint systems and the stable inversion. In order to clarify this relationship, we must extend the trial interval from the finite time interval [&lt;i&gt;T&lt;/i&gt;&lt;sub&gt;0&lt;/sub&gt;, &lt;i&gt;T&lt;sub&gt;f&lt;/sub&gt;&lt;/i&gt;] to the infinite time interval (-∞, ∞). However, when ILC is applied to actual systems, it is impossible to carry out the trial on an infinite time interval. So, the trial interval is necessarily truncated to the finite interval. Hence, we investigate the effect of the truncation on tracking performance.

DOI： 10.9746/sicetr1965.36.1117

researchmap

DOI： 10.9746/sicetr1965.36.1117

researchmap

記述言語：英語   掲載種別：研究論文（国際会議プロシーディングス）   出版者・発行元：IEEE  

Most of iterative learning control (ILC) using causal updating law obtains the input given by Silverman's or Hirshorn's causal inversion. When the objective system is that of a non-minimum phase, we cannot use those methods because the input is exponentially increasing. To overcome this difficulty, an approach called stable inversion was proposed to give a non-causal but bounded input instead. However, no simple iterative method to obtain this non-causal input was proposed. In this paper, from a viewpoint of minimization, we develop a simple iterative method for the stable inversion toward ILC for non-minimum phase systems.

Web of Science

researchmap

DOI： 10.5687/iscie.12.250

CiNii Books

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

掲載種別：研究発表ペーパー・要旨（全国大会，その他学術会議）  

J-GLOBAL

researchmap

記述言語：日本語   掲載種別：研究発表ペーパー・要旨（全国大会，その他学術会議）   出版者・発行元：電気学会  

CiNii Research

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

記述言語：日本語   出版者・発行元：一般社団法人電子情報通信学会  

画像中から直線を抽出することは物体認識を行う上で非常に重要な問題であり,中でも代表的な手法としてHough変換が知られている.しかし, Hough変換は直線として抽出されるパラメータ値から大きく離れたパラメータ値への冗長な投票が多く,処理時間に影響を与えてしまう.本稿では,画像空間における簡単な処理を用いることで,この冗長な投票を削減する手法を提案する.また,本手法が十分な精度を保ったまま高速化を実現するものであることを実験により確かめる.

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：画像電子学会  

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：一般社団法人映像情報メディア学会  

Although optical flow is useful for object tracking, estimation of the flow by the gradient method is ill-posed problem. In order to avoid ill-posedness, we have proposed a tracking method via 1D flow, which is calculated on the straight lines (calculation axes). We update these axes based on the positive 1D flow. Since, origin of the axes is located on the restricted position, we can't reduce a tracking error. We propose tracking method by using both 1D flow and frame difference.

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：一般社団法人映像情報メディア学会  

We propose a tracking algorithm based on mean shift procedure for occlusion. Mean shift tracking is rubust for partial occulusion. However, this method fails to track, when the most of the part of the object is occluded. We introduce a criteria of occulusion to rearrange the window if it is observed. Experimental result shows the efficiency of the proposed algorithm.

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：愛媛大学  

We consider the forward-propagation (FP) rule for an inverse kinematics problem. This method is based on the following steps. First, we construct an approximate inverse model. Second, the goal signal in each layer is derived by the Newton-like method. Finally, the updating of weights is realized by linear multiple regression. The regression coefficients express the adding correction of the weights. However, it isdifficult to obtain an accurate inverse kinematics model because of the structure of the network. Hence, we propose the FP rule with a high order term. The layer, which is constructed with the high order term of the input to the neural network, is inserted between the input layer and the hidden layer. We apply this rule to the inverse kinematics problem for the two-link manipulator and several desired trajectory. The simulation results show that the proposed method obtains a more accurate inverse kinematics model, compared to the FP rule without the high order term.

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：一般社団法人電子情報通信学会  

画像から直線を抽出する有効な手法としてハフ変換が知られているが,ハフ平面(ρ-θパラメータ空間)の離散化による抽出精度の問題がある.また,点群の配置が非常に疎で,ばらつきがある直線は,ハフ変換の性質上うまく抽出できないという問題もある.このようなハフ変換の苦手とする直線を適切に抽出する手法としてLMedSハフ変換が提案されている.LMedSハフ変換は直線の評価値としてLMedS値を用いており,ハフ変換の多数決原理とは異なったアルゴリズムにより直線を適切に近似して抽出することが可能である.ところが,LMedSハフ変換ではLMedS値の計算量が多いため,その処理時間に問題がある.そこで本稿では,LMedS値の計算量を削減することで高速に直線を抽出する手法を提案し,その有効性を実験により示す.

CiNii Books

researchmap

researchmap

記述言語：日本語   出版者・発行元：一般社団法人電子情報通信学会  

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：一般社団法人電子情報通信学会  

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：愛媛大学  

Estimation of the 2-D optical flow by the spatio-temporal method is ill-posed problem. Estimation of the 1-D optical flow is well-posed problem. However, a motion of the object on the image plane is limited in the linear direction along calculation axis of 1-D optical flow. In this paper, we propose a tracking method by spanning the several axes for calculating the 1-D optical flow. Our algorithm has the several parameters which have to be set up. We clarify the relation of the parameters and tracking error by experiment using test image sequence.

CiNii Books

researchmap

記述言語：日本語   出版者・発行元：一般社団法人電子情報通信学会  

直線や円などの画像構成要素(プリミティブ)を抽出する手法としてHough変換が知られている.筆者等は先に,このHough変換の投票配列の次元数を少なくすることで高速にプリミティブを抽出できる「ローカル極座標を用いたプリミティブの抽出法」を提案した.しかしながらこの方法には,「プリミティブの抽出速度が,画像に含まれる雑音の位置と量に影響を受ける」という性質(この性質を「雑音依存性」と呼ぶ)があり,これが実際にこの方法を利用する際の問題となっていた.そこで本橋では,この雑音依存性に対する解決策を提案する.提案する方法は,ローカル極の原点付近にウィンドウを設定し,このウィンドウ内の演算でローカル極設定の可否を判断するというもので,簡便な方法であるにも拘わらず高い効果を発揮することを種々の実験結果から明らかにしている.

CiNii Books

researchmap

開催年月日： 2022年

researchmap

開催年月日： 2022年

researchmap

開催年月日： 2019年12月

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

開催年月日： 2019年12月

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

開催年月日： 2019年9月

会議種別：口頭発表（一般）  

researchmap

開催年月日： 2019年9月

会議種別：口頭発表（一般）  

researchmap

開催年月日： 2019年9月

会議種別：口頭発表（一般）  

researchmap

開催年月日： 2019年9月

会議種別：口頭発表（一般）  

researchmap

開催年月日： 2019年9月

会議種別：口頭発表（一般）  

researchmap

開催年月日： 2019年8月

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：ポスター発表  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：ポスター発表  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

直線や円などの画像構成要素(プリミティブ)を抽出する手法としてHough変換が知られている、筆者等は先に,このHough変換の投票配列の次元数を少なくすることで高速にプリミティブを抽出できる「ローカル極座標を用いたプリミティブの抽出法」を提案した。しかしながらこの方法にはププリミティブの抽出速度が,画像に含まれる雑音の位置と量に影響を受ける」という性質(この性質を「雑音依存性」と呼ぶ)があり,これが実際にこの方法を利用する際の問題となっていた.そこで本橋では,この雑音依存性に対する解決策を提案する.提案する方法は,ローカル極の原点付近にウィンドウを設定し,このウィンドウ内の演算でローカル極設定の可否を判断するというもので,簡便な方法であるにも拘わらず高い効果を発揮することを種々の実験結果から明らかにしている.

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

画像中から直線を抽出することは物体認識を行う上で非常に重要な問題であり,中でも代表的な手法としてHough変換が知られている.しかし, Hough変換は直線として抽出されるパラメータ値から大きく離れたパラメータ値への冗長な投票が多く,処理時間に影響を与えてしまう.本稿では,画像空間における簡単な処理を用いることで,この冗長な投票を削減する手法を提案する.また,本手法が十分な精度を保ったまま高速化を実現するものであることを実験により確かめる.

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：ポスター発表  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

階層型ニューラルネットワーク上に逆モデルを構築する際の学習に，粒子群最適化を用いる．粒子群最適化で効果的な探索を行うには，群れの多様性を維持することが重要である．そこで，本研究では，群れの分裂と統合を繰り返すことで，多様性を維持する方法を提案する．

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：ポスター発表  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：ポスター発表  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

記述言語：日本語   会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

会議種別：口頭発表（一般）  

researchmap

記述言語：日本語  

画像中から直線を抽出することは物体認識を行う上で非常に重要な問題であり,中でも代表的な手法としてHough変換が知られている.しかし,Hough変換は直線として抽出されるパラメータ値から大きく離れたパラメータ値への冗長な投票が多く,処理時間に影響を与えてしまう.本稿では,画像空間における簡単な処理を用いることで,この冗長な投票を削減する手法を提案する.また,本手法が十分な精度を保ったまま高速化を実現するものであることを実験により確かめる.

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

researchmap

記述言語：日本語  

Although optical flow is useful for object tracking, estimation of the flow by the gradient method is ill-posed problem. In order to avoid ill-posedness, we have proposed a tracking method via 1D flow, which is calculated on the straight lines (calculation axes). We update these axes based on the positive 1D flow. Since, origin of the axes is located on the restricted position, we can't reduce a tracking error. We propose tracking method by using both 1D flow and frame difference.

researchmap

記述言語：日本語  

We propose a tracking algorithm based on mean shift procedure for occlusion. Mean shift tracking is rubust for partial occulusion. However, this method fails to track, when the most of the part of the object is occluded. We introduce a criteria of occulusion to rearrange the window if it is observed. Experimental result shows the efficiency of the proposed algorithm.

researchmap

記述言語：日本語  

画像から直線を抽出する有効な手法としてハフ変換が知られているが,ハフ平面(ρ-θパラメータ空間)の離散化による抽出精度の問題がある.また,点群の配置が非常に疎で,ばらつきがある直線は,ハフ変換の性質上うまく抽出できないという問題もある.このようなハフ変換の苦手とする直線を適切に抽出する手法としてLMedSハフ変換が提案されている.LMedSハフ変換は直線の評価値としてLMedS値を用いており,ハフ変換の多数決原理とは異なったアルゴリズムにより直線を適切に近似して抽出することが可能である.ところが,LMedSハフ変換ではLMedS値の計算量が多いため,その処理時間に問題がある.そこで本稿では,LMedS値の計算量を削減することで高速に直線を抽出する手法を提案し,その有効性を実験により示す.

researchmap