From the Publisher: 
The accessible presentation of this book gives both a general view of the entire computer vision enterprise and also offers sufficient detail to be able to build useful applications. Users learn techniques that have proven to be useful by first-hand experience and a wide range of mathematical methods. A CD-ROM with every copy of the text contains source code for programming practice, color images, and illustrative movies. Comprehensive and up-to-date, this book includes essential topics that either reflect practical significance or are of theoretical importance. Topics are discussed in substantial and increasing depth. Application surveys describe numerous important application areas such as image based rendering and digital libraries. Many important algorithms broken down and illustrated in pseudo code. Appropriate for use by engineers as a comprehensive reference to the computer vision enterprise.

Computer vision : a modern approach = 计算机视觉 : 一种现代的方法

Visual perception refers to the process of organizing, identifying, and interpreting visual information in environmental awareness and understanding. With the rapid progress of multimedia acquisition technology, research on visual perception has been a hot topic in the academical field and industrial applications. Especially after the introduction of artificial intelligence theory, intelligent visual perception has been widely used to promote the development of industrial production towards intelligence. In this article, we review the previous research and application of visual perception in different industrial fields such as product surface defect detection, intelligent agricultural production, intelligent driving, image synthesis, and event reconstruction. The applications basically cover most of the intelligent visual perception processing technologies. Through this survey, it will provide a comprehensive reference for research on this direction. Finally, this article also summarizes the current challenges of visual perception and predicts its future development trends.

/pdf/visual-perception-enabled-industry-intelligence-state-of-the-3v5dsw2daa.pdf

Visual Perception Enabled Industry Intelligence: State of the Art, Challenges and Prospects

GS2-2 10.5954/ICAROB.2021.GS2-2 https://www.doi.org/10.5954/ICAROB.2021.GS2-2 Image Registration Method for Chest MDCT Images Based on 2-D Finite Element Method GS2-3 10.5954/ICAROB.2021.GS2-3 https://www.doi.org/10.5954/ICAROB.2021.GS2-3 Detection of Abnormal Shadows in Low-dose CT Images Using CNN GS2-4 10.5954/ICAROB.2021.GS2-4 https://www.doi.org/10.5954/ICAROB.2021.GS2-4 EEG Signal Extraction Method Based on HHT and CSP

A Method for Improving Recognition of Lying Postures Using a Measured Signal Intensity of Respiration and Heartbeat by Flexible Tactile Sensor Sheet

Standard industrial robot manipulator has possibility to manipulate the objects accurately in arbitrary position by using additional vision sensor. This applies in eye-to-hand camera configuration, where the stereo camera can cover the robot arm workspace and the object can be kept accurately for three dimension position estimation. The goal of this system is to control a robot arm to pick up and place the targeted object autonomously within its workspace. In order to accomplish this goal, a stereo vision technique including eye to hand calibration is applied. First, a stereo camera is calibrated to obtain the intrinsic and extrinsic camera parameters. Then, the specific color features of desired object are extracted using image processing algorithms. After that, based on centroid of desired object, 3D (3 dimension) position of the desired object is calculated. Then, camera and robot arm coordinate transformation was calibrated using adaptive artificial network based fuzzy interference system (ANFIS) method. Finally, the 3D coordinates of desired object are sent to robot arm controller as the inputs to move the end-effector equipped with gripper to the object location and pick up the object then place it to the desired location. The experiment was carried out using 6 DOF industrial robot arm and the results demonstrated the good performance of the system with acceptable errors.

Eye to hand calibration using ANFIS for stereo vision-based object manipulation system

A method and an apparatus for generating an image with shallow depth of field, suitable for a three-dimensional imaging system having a left lens and a right lens, are provided. First, the left lens and the right lens are used to capture a left-eye image and a right-eye image. Next, a disparity between each of a plurality of corresponding pixels in the left and right-eye images is calculated. Then, the depth information of each pixel is estimated according to a focus of the left and right lens, a lens pitch between the left lens and the right lens, and the calculated disparity of each pixel. Finally, the pixels in the left-eye image and the right-eye image are blurred according to the focus of the left lens and the right lens and the estimated depth information of each pixel, so as to obtain the image with shallow depth of field.

Method and apparatus for generating image with shallow depth of field

This system is modeled on the distance perception methods of human eyes. It uses DSP to process the images obtained by two CCD cameras and calculate the distance by the principle of parallax. This paper describes the working principle and hardware platform design. It focuses on the image matching methods based on Canny edge detection and Harris corner detection. The results of the experiment show that the system is efficient and reliable.

Distance Measurement System Based on Binocular Stereo Vision

In recent years, the research of intelligent transportation system has become more and more popular. Whether it was the environmental information around the vehicle, the landmarks on the road or the position of the vehicle itself. It played a very important role in the intelligent traffic control of autonomous mobile intelligent vehicle or vehicle driving assistance system. This paper analyzes the 3D point cloud data collecting and constructing by the radar to obtain the location and distance between the self-propelled vehicle and the road marking in the real environment space. In the part of detecting the road edge, it uses the characteristics of the height difference between the driving road and the sidewalk or fence, as well as the distance between two points in point cloud data and gradient filter to divide road area and non-Road area. In order to obtain more complete and accurate environmental data, the non-road data segmented uses to splice the feature points of the continuous picture through the previous method, and put calculated conversion matrix into the road area to obtain two kinds of information: the non-driving road environment map and the feasible driving road area map. In the 3D map information of drivable pavement area, the lane markings may be extracted from the asphalt pavement and road markings according to the different of LiDAR reflection intensity. Then the road edge detected on the point cloud image is projected onto the 2D color image. The detection and tracking of lane markings on the road area image are realized by image processing method, and the experiment results show that the marking machine or self-propelled vehicle can move along the lane marking accurately.

Road perception and road line detection based on fusion of LiDAR and camera

This paper uses the point cloud data extracted by LiDAR as the basis of obstacle detection and proposes an algorithm to reduce the computational complexity of point cloud. Firstly, it removes the points too high or too low from the point cloud data, leaving only the points forming the obstacle interval. Secondly, the point cloud in the interval is projected to the X-Y plane (horizontal plane) and the resolution is reduced by lattice. Finally, the connected component marker clustering method is used to segment and classify obstacles. After identifying the obstacle, the scene information is provided to the path planning algorithm. In the path planning algorithm, Unmanned Ground Vehicle (UGV) moves along the shortest straight line connecting the target point and the starting point. When encountering obstacles, the edge tracking method is used to bypass the obstacle, and then it goes straight again. After planning by the path planning algorithm, UGV can avoid collision and advance to the destination in real time through the motion control system. The experimental result shows that UGV designed by this research has the ability to reach the target point safely, and it can avoid obstacles and move autonomously in real-time conditions and complete outdoor unknown environmental obstacle detection.

UGV obstacle detection of sensor fusion based on point cloud data segmentation

Simultaneous Localization and Mapping (SLAM) is a technology developed to overcome the problems of robot localization and navigation. This paper uses a simple and low-cost way to establish indoor environment information map data for work planning and patrol monitoring of self-propelled robots. The advanced mobile robot uses mainly Lidar and camera to sense the environment. Being of its high accuracy and large sensing range, the robot gets a large amount of environment sensing data, use it on SLAM to obtain a detailed environment map, and locate the robot location. However, in order for the robot to use Lidar and camera to detect environmental data, it must use higher-level controller and memory to deal with a large number of sensing data, resulting in an increase in the price of sensors and the cost of matching hardware. In this paper, a low-cost single line Lidar, combined with DC motor and rotary encoder, is installed on a small robot, so that it obtains the data of different measurement points, and the robot is used as the test vehicle to explore the indoor environment. The experimental result shows that its sensing characteristics are sensitive, and the sensing data meet the requirements of real-time mapping.

Tianding Chen

Papers

Distance Measurement System Based on Binocular Stereo Vision

Road perception and road line detection based on fusion of LiDAR and camera

UGV obstacle detection of sensor fusion based on point cloud data segmentation

A fast mapping for small mobile robot scanned by single line Lidar