人工智能与计算机视觉产业发展

随着世界科技化进程的不断加快,人工智能领域逐渐被推向世界舞台,作为人工智能领域中的重要一部分,计算机视觉技术已取得了较大的进步。计算机视觉发展的目标是为了能够让机器和人类一样拥有视觉,进行事物识别,计算机视觉已从之前对计算机图片的识别慢慢转变为对实际生活中的识别,由此迈出了技术发展的重要一步。但计算机视觉在发展过程中也不是一帆风顺的,在发展中也存在一些问题,笔者就人工智能与计算机视觉领域的发展这一话题进行深入探讨。     

基于计算机视觉的运动目标分析

在计算机里面,所谓的计算机视觉的运动目标分析,就是对视频里面的运动目标进行跟踪检测,并对目标进行研究分析,而检测与跟踪是对分析目标和理解目标行为的基础,因此需要重视其中的目标.本文通过计算机视觉分析了跟踪与检测目标等问题,从而为检测和和跟踪视觉运动目标的算法做出重大意义.     

深度学习在目标视觉检测中的应用进展与展望

目标视觉检测是计算机视觉领域的一个重要问题,在视频监控、自主驾驶、人机交互等方面具有重要的研究意义和应用价值.近年来,深度学习在图像分类研究中取得了突破性进展,也带动着目标视觉检测取得突飞猛进的发展.本文综述了深度学习在目标视觉检测中的应用进展与展望.首先对目标视觉检测的基本流程进行总结,并介绍了目标视觉检测研究常用的公共数据集;然后重点介绍了目前发展迅猛的深度学习方法在目标视觉检测中的最新应用进展;最后讨论了深度学习方法应用于目标视觉检测时存在的困难和挑战,并对今后的发展趋势进行展望.     

用于高精度三维计算机视觉的图象系统标定和误差补偿

为了对计算机视觉系统进行标定,并削弱由系统模型与真实系统间差异所产生的误差,在详细分析计算机数字成象过程的基础上,提出了一种综合数字成象过程中内部的及外部的,已知的及未知的等各种因素的图象系统标定方法,其标定的具体对象包括光学成象过程中的线性几何投影参数及非线性畸变参数、光电转换过程中的参数和图象信号的D／A及A／D转换过程中的参数等;同时提出了基于该标定方法的三维计算机视觉系统采集三维数据的误差补偿方法及其参数的确定过程。上述方法已经成功地应用于三维计算机视觉系统,并取得了很好的三维精度。     

光场成像技术及其在计算机视觉中的应用

目的 光场成像技术刚刚在计算机视觉研究中展开初步应用,其相关研究比较零散,缺乏系统性。本文旨在系统介绍光场成像技术发展以及其应用在计算机视觉研究中有代表性的工作。方法 从解决计算机视觉问题的角度出发,4个层面讨论光场成像技术最近十年的研究工作,包括:1)主流的光场成像设备及其作为计算机视觉传感器的优点与不足;2)光场相机作为视觉传感器的标定、解码以及预处理方法;3)基于4维光场的图像渲染与重建技术,以及其如何促进计算机视觉研究;4)以4维光场数据为基础的特征表达方法。结果 逐层梳理出光场成像在求解视觉问题中的优势和局限,分析其中根本性的原理与掣肘,力图总结出亟待解决的关键问题以及未来的发展趋势。结论 作为一种颇具前景的新型计算机视觉传感器技术,光场成像技术的研究必将更为广泛和深入。研究应用于计算机视觉的光场成像技术将有力的引导和促进计算机视觉和光场成像技术协同发展。     

机器视觉系统中的照明优化设计

在设计机器视觉系统的照明部分时,需要根据应用环境和设备成本来选择光源,涉及到光源的照射方式、色谱范围、照明类型、光波极性、光强等问题。为了使被拍摄物能够在相机传感器上清晰成像,需要分析从被拍摄物反射到相机传感器上的光能及所需光源能量。结合CCD特性,提出了一种照明优化设计方法,导出了CCD传感器的成像值和光照能量之间的关系,对机器视觉系统中合理选择光源和优化照明设计具有指导作用,在检测技术与自动化装置研究领域具有重要的实用意义。     

多传感器信息融合的服务机器人导航方法

在机器人中,计算机视觉与传感器开始交融.本文设计了以人脸识别为视觉导引的主体,传感器为辅的服务机器人系统,提出了一种基于视觉的图像处理算法,采用将形态学处理与自适应阈值分割相结合直接除掉阴影,使用多传感器与视觉融合的方法来解决视觉误判的问题.在小车避障中运用单目测距的原理,跳过了传统相机的标定问题,并将该算法与迷宫机器人路径规划中的左手法则相结合运用在设计的服务机器人中.     

3D目标检测进展综述

目标检测算法应用广泛,一直是计算机视觉领域备受关注的研究热点。近年来,随着深度学习的发展,3D图像的目标检测研究取得了巨大的突破。与2D目标检测相比,3D目标检测结合了深度信息,能够提供目标的位置、方向和大小等空间场景信息,在自动驾驶和机器人领域发展迅速。文中首先对基于深度学习的2D目标检测算法进行概述;其次根据图像、激光雷达、多传感器等不同数据采集方式,分析目前具有代表性和开创性的3D目标检测算法;结合自动驾驶的应用场景,对比分析不同3D目标检测算法的性能、优势和局限性;最后总结了3D目标检测的应用意义以及待解决的问题,并对3D目标检测的发展方向和新的挑战进行了讨论和展望。     

计算机视觉下的果实目标检测算法综述

基于计算机视觉的果实目标检测识别是目标检测、计算机视觉、农业机器人等多学科的重要交叉研究课题,在智慧农业、农业现代化、自动采摘机器人等领域,具有重要的理论研究意义和实际应用价值。随着深度学习在图像处理领域中广泛应用并取得良好效果,计算机视觉技术结合深度学习方法的果实目标检测识别算法逐渐成为主流。本文介绍基于计算机视觉的果实目标检测识别的任务、难点和发展现状,以及2类基于深度学习方法的果实目标检测识别算法,最后介绍用于算法模型训练学习的公开数据集与评价模型性能的评价指标,且对当前果实目标检测识别存在的问题和未来可能的发展方向进行讨论。     

基于计算机视觉的烟叶自动分级系统硬件设计

介绍了一种基于计算机视觉烟叶自动分级系统的硬件组成,该系统由输送装置、称重系统、视觉系统、均匀照明室4部分组成。输送装置可将烟叶打散摊平,使烟叶保持一定的厚度呈现给CCD摄像头;称重系统采用电子皮带秤完成烟叶的在线实时称重,并将数据发送至计算机内存;视觉系统同时监视、采集烟叶图像信息送至计算机内存进行处理;照明系统为CCD摄像头均匀、恒定的光源,保证图像的清晰。实验证明了系统的可行性,其辅助机构与烟叶实时分级决策算法尚待进一步的研究。     

Illumination setup planning for a hand-eye system based on an environmental model

In hand-eye systems for advanced robotic applications such as assembly, the degrees of freedom of the vision sensor should be increased and actively made use of to cope with unstable scene conditions. Particularly, in the case of using a simple vision sensor, an intelligent adaptation of the sensor is essential to compensate for its inability to adapt to a changing environment. This paper proposes a vision sensor setup planning system which operates based on environmental models and generates plans for using the sensor and its illumination assuming freedom of positioning for both. A typical vision task in which the edges of an object are measured to determine its position and orientation is assumed for the sensor setup planning. In this context, the system is able to generate plans for the camera and illumination position, and to select a set of edges best suited for determining the object's position. The system operates for stationary or moving objects by evaluating scene conditions such as edge length, contrast, and relative angles based on a model of the object and the task environment. Automatic vision sensor setup planning functions, as shown in this paper, will play an important role not only for autonomous robotic systems, but also for teleoperation systems in assisting advanced tasks.     

Object oriented concepts and mechanisms for feature-based computer integrated inspection

Computer vision can provide fast, accurate and reliable inspection. In this paper is described a knowledge-based vision inspection planning system for manufactured components based on their CAD design models. An object oriented methodology is developed to capture the knowledge such as the geometric entities, their relationships, sensors and plans. Reasoning mechanisms are developed to automatically determine various attributes of the different features of an object to be inspected, and to generate the alternative strategies for inspection of each attribute. Mechanisms for automated generation of sensor settings and optimization of the inspection plan are also developed. Sensor constraints are considered in determining the camera parameters. This inspection methodology is flexible, and it is easily integrated with the design and process planning aspects of an integrated manufacturing system.     

Fusing force and vision feedback for manipulating deformable objects

This article describes a framework that fuses vision and force feedback for the control of highly deformable objects. Deformable active contours, or snakes, are used to visually observe changes in object shape over time. Finite‐element models of object deformations are used with force feedback to predict expected visually observed deformations. Our approach improves the performance of large, complex deformable contour tracking over traditional computer vision tracking techniques. This same approach of combining deformable active contours with finite‐element material models is modified so that a vision sensor, i.e., a charge‐coupled device (CCD) camera, can be used as a force sensor. By visually tracking changes in contours on the object, material deflections can be transformed into applied stress estimates through finite element modeling. Therefore, internal object stresses due to object manipulation can be visually observed and controlled, thus creating a framework for deformable object manipulation. A pinhole camera model is used to accomplish vision and force sensor feedback assimilation from these two sensing modalities during manipulation. © 2001 John Wiley & Sons, Inc.     

A low-complexity sensor fusion algorithm based on a fiber-optic gyroscope aided camera pose estimation system

Visual tracking, as a popular computer vision technique, has a wide range of applications, such as camera pose estimation. Conventional methods for it are mostly based on vision only, which are complex for image processing due to the use of only one sensor. This paper proposes a novel sensor fusion algorithm fusing the data from the camera and the fiber-optic gyroscope. In this system, the camera acquires images and detects the object directly at the beginning of each tracking stage; while the relative motion between the camera and the object measured by the fiber-optic gyroscope can track the object coordinate so that it can improve the effectiveness of visual tracking. Therefore, the sensor fusion algorithm presented based on the tracking system can overcome the drawbacks of the two sensors and take advantage of the sensor fusion to track the object accurately. In addition, the computational complexity of our proposed algorithm is obviously lower compared with the existing approaches(86% reducing for a 0.5 min visual tracking). Experiment results show that this visual tracking system reduces the tracking error by 6.15% comparing with the conventional vision-only tracking scheme(edge detection), and our proposed sensor fusion algorithm can achieve a long-term tracking with the help of bias drift suppression calibration.     

Tracking in 3D: Image Variability Decomposition for Recovering Object Pose and Illumination

As an object moves through space, it changes its orientation relative to the viewing camera and relative to light sources which illuminate it. As a consequence, the images of the object produced by the viewing camera may change dramatically. Thus, to successfully track a moving object, image changes due to varying pose and illumination must be accounted for. In this paper, we develop a method for object tracking that can not only accommodate large changes in object pose and illumination, but can recover these parameters as well. To do this, we separately model the image variation of the object produced by changes in pose and illumination. To track the object through each image in the sequences, we then locally search the models to find the best match, recovering the object’s orientation and illumination in the process. Throughout, we present experimental results, achieved in real-time, demonstrating the effectiveness of our methods. Received: 9 December 1998?Received in revised form: 23 December 1998?Accepted: 23 December 1998     

由特征点的空间位置估计运动参数

确定被观察物体和摄像机之间的三维相对运动是计算机视觉领域的一个重要课题.假定物体上有若干特征点,它们在空间的位置已由双目视觉求得.在此假定下,本文讨论如何由特征点空间位置的观察值来估计运动参数.现有文献中所用的目标函数导致的是一个非线性最优化问题.本文利用凯莱(Cayley)定理提出一种新的目标函数,它导致的是一个十分简单的线性最小二乘问题.文中还讨论了这两种目标函数之间的关系.     

3D model based tracking for omnidirectional vision: A new spherical approach

The current work addresses the problem of 3D model tracking in the context of monocular and stereo omnidirectional vision in order to estimate the camera pose. To this end, we track 3D objects modeled by line segments because the straight line feature is often used to model the environment. Indeed, we are interested in mobile robot navigation using omnidirectional vision in structured environments. In the case of omnidirectional vision, 3D straight lines are projected as conics in omnidirectional images. Under certain conditions, these conics may have singularities.In this paper, we present two contributions. We, first, propose a new spherical formulation of the pose estimation withdrawing singularities, using an object model composed of lines. The theoretical formulation and the validation on synthetic images thus show that the new formulation clearly outperforms the former image plane one. The second contribution is the extension of the spherical representation to the stereovision case. We consider in the paper a sensor which combines a camera and four mirrors. Results in various situations show the robustness to illumination changes and local mistracking. As a final result, the proposed new stereo spherical formulation allows us to localize online a robot indoor and outdoor whereas the classical formulation fails.     

A Catadioptric and Pan-Tilt-Zoom Camera Pair Object Tracking System for UAVs

Unmanned aerial vehicles (UAVs) are seeing widespread use in military, scientific, and civilian sectors in recent years. As the mission demands increase, these systems are becoming more complicated. Omnidirectional camera is a vision sensor that can captures 360° view in a single frame. In recent years omnidirectional camera usage has experienced a remarkable increase in many fields, where many innovative research has been done. Although, it is very promising, employment of omnidirectional cameras in UAVs is quite new. In this paper, an innovative sensory system is proposed, that has an omnidirectional imaging device and a pan tilt zoom (PTZ) camera. Such a system combines the advantages of both of the camera systems. The system can track any moving object within its 360° field of view and provide detailed images of it. The detection of the moving object has been accomplished by an adaptive background subtraction method implemented on the lowered resolution images of the catadioptric camera. A novel algorithm has also been developed to estimate the relative distance of the object with respect to the UAV, using tracking information of both of the cameras. The algorithms are implemented on an experimental system to validate the approach.     

Object matching in disjoint cameras using a color transfer approach

Object appearance models are a consequence of illumination, viewing direction, camera intrinsics, and other conditions that are specific to a particular camera. As a result, a model acquired in one view is often inappropriate for use in other viewpoints. In this work we treat this appearance model distortion between two non-overlapping cameras as one in which some unknown color transfer function warps a known appearance model from one view to another. We demonstrate how to recover this function in the case where the distortion function is approximated as general affine and object appearance is represented as a mixture of Gaussians. Appearance models are brought into correspondence by searching for a bijection function that best minimizes an entropic metric for model dissimilarity. These correspondences lead to a solution for the transfer function that brings the parameters of the models into alignment in the UV chromaticity plane. Finally, a set of these transfer functions acquired from a collection of object pairs are generalized to a single camera-pair-specific transfer function via robust fitting. We demonstrate the method in the context of a video surveillance network and show that recognition of subjects in disjoint views can be significantly improved using the new color transfer approach.     

多目标背景下双目视觉深度信息优化研究及其实现

针对机器人领域应用视觉进行目标物体抓取问题,提出了一种针对多目标背景下,新的深度优化处理方法.通过设定一个阈值块,以遍历成块的深度信息用类似聚类的方法,提出目标物体的具体坐标,传递给机器人手臂,完成准确的抓取操作.依次介绍了双目视觉原理、摄像机标定、双目矫正和双目匹配等内容,以及呈现出原始的深度信息图以及优化后的深度信息图,比较它们的差距.最后在实验中给出了证明:此种深度信息优化方法能够有效的提高机器人抓取目标物体的成功率.最后,还在文章最后给出了下一步的研究方向.