Current Issue : January - March Volume : 2020 Issue Number : 1 Articles : 5 Articles
In the printing industry, defect detection is of crucial importance for ensuring the quality\nof printed matter. However, rarely has research been conducted for web offset printing. In this\npaper, we propose an automatic defect detection method for web offset printing, which consists of\ndetermining first row of captured images, image registration and defect detection. Determining\nthe first row of captured images is a particular problem of web offset printing, which has not been\nstudied before. To solve this problem, a fast computational algorithm based on image projection\nis given, which can convert 2D image searching into 1D feature matching. For image registration,\na shape context descriptor is constructed by considering the shape concave-convex feature, which can\neffectively reduce the dimension of features compared with the traditional image registration method.\nTo tolerate the position difference and brightness deviation between the detected image and the\nreference image, a modified image subtraction is proposed for defect detection. The experimental\nresults demonstrate the effectiveness of the proposed method....
This paper presents the design and construction of a robotic arm that plays chess against a\nhuman opponent, based on an artificial vision system. The mechanical design was an adaptation of\nthe robotic arm proposed by the rapid prototyping laboratory FabLab RUC (Fabrication Laboratory\nof the University of Roskilde). Using the software Solidworks, a gripper with 4 joints was designed.\nAn artificial vision system was developed for detecting the corners of the squares on a chessboard and\nperforming image segmentation. Then, an image recognition model was trained using convolutional\nneural networks to detect the movements of pieces on the board. An image-based visual servoing\nsystem was designed using the Kanadeâ??Lucasâ??Tomasi method, in order to locate the manipulator.\nAdditionally, an Arduino development board was programmed to control and receive information\nfrom the robotic arm using Gcode commands. Results show that with the Stockfish chess game\nengine, the system is able to make game decisions and manipulate the pieces on the board. In this\nway, it was possible to implement a didactic robotic arm as a relevant application in data processing\nand decision-making for programmable automatons....
In recent years, Convolutional Neural Networks (CNNs) have enabled unprecedented\nprogress on a wide range of computer vision tasks. However,\ntraining large CNNs is a resource-intensive task that requires specialized\nGraphical Processing Units (GPU) and highly optimized implementations to\nget optimal performance from the hardware. GPU memory is a major bottleneck\nof the CNN training procedure, limiting the size of both inputs and\nmodel architectures. In this paper, we propose to alleviate this memory bottleneck\nby leveraging an under-utilized resource of modern systems: the device\nto host bandwidth. Our method, termed CPU offloading, works by\ntransferring hidden activations to the CPU upon computation, in order to\nfree GPU memory for upstream layer computations during the forward pass.\nThese activations are then transferred back to the GPU as needed by the gradient\ncomputations of the backward pass. The key challenge to our method is\nto efficiently overlap data transfers and computations in order to minimize\nwall time overheads induced by the additional data transfers. On a typical\nwork station with a Nvidia Titan X GPU, we show that our method compares\nfavorably to gradient checkpointing as we are able to reduce the memory\nconsumption of training a VGG19 model by 35% with a minimal additional\nwall time overhead of 21%. Further experiments detail the impact of the different\noptimization tricks we propose. Our method is orthogonal to other\ntechniques for memory reduction such as quantization and sparsification so\nthat they can easily be combined for further optimizations....
With the rapid development of machine vision, binocular stereo vision based\non the principle of parallax has gradually become the core of scientific research.\nThis paper briefly presents the background and research significance,\nelaborates the research status of binocular vision robot at home and abroad\nand studies the checkerboard calibration method, and uses Matlab to complete\nbinocular camera calibration. Stereo matching technology is the core\nand most difficult part of binocular stereoscopic 3D reconstruction research.\nFirstly, the image acquired after calibration is enhanced by gray scale transformation\nto make the image clearness optimal, and then use NCC (normalization\ncross-compilation). The algorithm performs the matching of left and\nright image pairs in the Matlab environment to generate an optimal matching\ndisparity map....
Visible light and infrared bands of the optical spectrum used for optical camera\ncommunication (OCC) are becoming a promising technology nowadays. Researchers are proposing\nnew OCC-based architectures and applications in both indoor and outdoor systems using the\nembedded cameras on smartphones, with a view to making them user-friendly. Smartphones have\nuseful features for developing applications using the complementary metal-oxide-semiconductor\ncameras, which can receive data from optical transmitters. However, several challenges have arisen\nin increasing the capacity and communication range, owing to the limitations of current cameras\nand implementation complexities. In this paper, we provide a comprehensive analysis of the OCC\ntechnology requirements and opportunities using smartphone cameras from an implementation\npoint of view. Furthermore, we demonstrate an OCC system using a low frame rate smartphone\ncamera to particularly analyze the requirements and critical implementation challenges. Also, some\npossible solutions are provided with a view to improving the overall system capacity, communication\ndistance, and stability....
Loading....