In surveillance systems a constant monitoring is required for high security purpose, however the videos or the images captured here can be degraded because of low illumination environment and foggy atmosphere. In this...
详细信息
ISBN:
(纸本)9781538640081
In surveillance systems a constant monitoring is required for high security purpose, however the videos or the images captured here can be degraded because of low illumination environment and foggy atmosphere. In this paper we have developed a system that acquires images through motion detection and enhance the quality of image with illumination adjustment and haze removal algorithms in FPGA. The motion detection algorithm has been implemented in MATLAB r2013a whereas the acquired image has been processed in FPGA Virtex 6 ML605 evaluation board. The visual quality of the image cannot be judged merely by observing image but also by determining some parameters like PSNR, MSE etc. So on the basis of implementation results it has been observed that the illumination levels of the image has been adjusted and the haze has been removed to some extent and in addition to that the system requires less time for processing purpose.
The parallelization of programs and distributing their workloads to multiple threads can be a challenging task. In addition to multithreading, harnessing vector units in CPUs proves highly desirable. However, employin...
详细信息
ISBN:
(纸本)9781450350303
The parallelization of programs and distributing their workloads to multiple threads can be a challenging task. In addition to multithreading, harnessing vector units in CPUs proves highly desirable. However, employing vector units to speed up programs can be quite tedious. Either a program developer solely relies on the auto-vectorization capabilities of the compiler or he manually applies vector intrinsics, which is extremely error-prone, difficult to maintain, and not portable at all. Based on whole-function vectorization, a method to replace control flow with data flow, we propose auto-vectorization techniques for imageprocessing DSLs in the context of source-to-source compilation. The approach does not require the input to be available in SSA form. Moreover, we formulate constraints under which the vectorization analysis and code transformations may be greatly simplified in the context of imageprocessing DSLs. As part of our methodology, we present control flow to data flow transformation as a source-to-source translation. Moreover, we propose a method to efficiently analyze algorithms with mixed bit-width data types to determine the optimal SIMD width, independently of the target instruction set. The techniques are integrated into an open source DSL framework. Subsequently, the vectorization capabilities are compared to a variety of existing state-of-the-art C/C++ compilers. A geometric mean speedup of up to 3.14 is observed for benchmarks taken from ISPC and imageprocessing, compared to non-vectorized executions.
A novel patch-based multi-view image denoising algorithm is proposed. This method leverages the 3D focus image stacks structure to exploit self-similarity and image redundancy inherent in multiple view images. Then a ...
详细信息
ISBN:
(纸本)9781509041176
A novel patch-based multi-view image denoising algorithm is proposed. This method leverages the 3D focus image stacks structure to exploit self-similarity and image redundancy inherent in multiple view images. Then a depth-guided adaptive window and dynamic view selection criterion is developed to aid proper selection of most consistent patches for the multi-view image denoising. Extensive experiments have been performed. Comparing the outcomes against those of state of the art image denoising algorithms, our proposed algorithm demonstrates significant performance advantage.
The transformation of the development of the Arctic is due to modern robotic systems. The use of unmanned vehicles in many industries in the Arctic provides an array of photo and video information. Accuracy of image a...
The transformation of the development of the Arctic is due to modern robotic systems. The use of unmanned vehicles in many industries in the Arctic provides an array of photo and video information. Accuracy of image analysis and pattern recognition is enhanced by image preprocessing. However, the existing binarization algorithms are not universal for images with different distortions and loss of information. The accuracy of binarization algorithms depends on many factors, such as shadows, uneven lighting, low contrast, noise, etc. images with different characteristics of light and noise are simulated in order to model various lighting conditions on information from digital cameras of robotic systems. The paper investigates global and adaptive image binarization algorithms. The binarized images were obtained using these algorithms and the results of binarized images recognition are compared by an optical character recognition system. An analysis of the comparison results showed that for images made in poor lighting conditions or images with low contrast, or images with high noise levels, adaptive binarization algorithms are better suited. However, in most cases it is not possible to obtain fully correctly recognized *** paper proposes a new binarization method based on artificial neural networks. The process of creating an artificial neural network is shown, include the parameters for determining the class of a pixel, the adjustment of weights, the architecture of an artificial neural network. A comparison of the proposed artificial neural network with existing image binarization algorithms demonstrate that in most cases the artificial neural network has the result of imageprocessing at the level of adaptive algorithms or higher. The proposed method of images binarization based on the image color characteristics analysis allows to solve image recognition tasks by robotized systems.
It is important to improve the integrity and accuracy of sonar image target detection, which is significant for underwater detection. In this paper, a variety of sonar image denoising algorithms and segmentation algor...
详细信息
ISBN:
(纸本)9781538635735
It is important to improve the integrity and accuracy of sonar image target detection, which is significant for underwater detection. In this paper, a variety of sonar image denoising algorithms and segmentation algorithms are studied, and a denoising algorithm based on fast curve transform is proposed. The image segmentation algorithm based on k-means clustering is studied, and the optimal clustering number screening and sonar image subsurface segmentation are realized. The sonar image fast segmentation algorithm based on ICM algorithm and the object contour detection of sonar image based on level set method are realized in Matlab. The results show that the proposed algorithm can improve the noise reduction effect of the sonar image under reverberation interference, and obtain a better image detection effect.
This work presents a monitoring system for tactical forest fire-fighting operations based on a team of unmanned aerial vehicles and remote sensing techniques. Functions and missions of the system, as well as its archi...
详细信息
ISBN:
(纸本)9781538663844
This work presents a monitoring system for tactical forest fire-fighting operations based on a team of unmanned aerial vehicles and remote sensing techniques. Functions and missions of the system, as well as its architecture, are considered. imageprocessing and remote sensing algorithms are presented, a way for data integration into a fire-spreading model in a real-time forest fire response decision support system is proposed. The combination of automatic monitoring and remote sensing techniques with an approximate fire-spreading model can provide required credibility and efficiency of fire prediction and response.
The flame is a visually unstable and constantly changeable process, which causes considerable difficulties for its detection in the video streams. Although the modern architecture of convolutional neural networks can ...
详细信息
ISBN:
(纸本)9781538628751
The flame is a visually unstable and constantly changeable process, which causes considerable difficulties for its detection in the video streams. Although the modern architecture of convolutional neural networks can show high accuracy, their integration into real-time systems is problematic, because they require a large amount of computing resources. To reduce the number of these resources, it is proposed to select possible regions of interest (ROI), which are based on the developed generator of hypotheses. Compared to existing flame detection algorithms, the developed generator of hypotheses allows you to work with the minimum of computing resources and has a high degree of classification completeness due to improved methods of color segmentation and moving objects detection.
Convolutional neural networks (CNNs) have a wide range of applications in image and video recognition, recommender systems and natural language processing. But CNNs are computationally intensive, and its computational...
详细信息
ISBN:
(纸本)9781538683682
Convolutional neural networks (CNNs) have a wide range of applications in image and video recognition, recommender systems and natural language processing. But CNNs are computationally intensive, and its computational cost is hard to accept. In order to speed up the calculations, people focus on optimizing convolution that account for most of the proportion of CNNs' operation. So, many algorithms have been proposed to accelerate the operation of convolution layers. However, each algorithm has its advantages and disadvantages, and there is no one algorithm that can handle all situations. In this paper, we examine the performance of various algorithms in GPU environment. By building a customized CNN model, we have fully explored the impact of the neural structure on the performance of algorithms, including inference/training speed, memory consumption and power consumption. In addition to the algorithms, we also focus on how their implementations in GPU environment affect their performance. We trace the kernel functions of these implementations to further generalize the characteristics of these algorithms. Finally, we summarize the characteristics of each algorithm., and design a strategy to assigns the appropriate implementation for different convolutional layers in CNNs. With our strategy, we can make AlexNet run 1.2× to 2.8× faster than other strategies in GPU environment. This work has very important meaning for understanding these algorithms and may provide insights for further optimizations of the architecture of GPUs and accelerators.
with the rapid advancement in the internet, we are now living in the era of big data. The image data over the web has the potential to assist in the development of sophisticated and robust models and algorithms to int...
详细信息
ISBN:
(纸本)9781538626344;9781538626337
with the rapid advancement in the internet, we are now living in the era of big data. The image data over the web has the potential to assist in the development of sophisticated and robust models and algorithms to interact with images and multimedia data. images Data sets are widely used in imageprocessing tasks and analyses. They are in use in various fields including Artificial Intelligence, Data extraction and collection, Computer Vision, Research studies and education. In this research work, we are going to propose a system that crawls the web in a systematic manner using Hadoop MapReduce technique to collect images from millions of web pages on the web. With Celebrity images just a use case, we would be able to search for and retrieve any image tagged with some specific terms. The system uses some simple techniques to reduce noisy images like thumbnails and icons. The proposed system is based on Apache Hadoop and Apache Nutch, an open source web crawler. A customized crawl is run through Apache Nutch in a Hadoop Cluster that searches images for one or more categories on the web and retrieves their links. Next, HIPI, Hadoop imageprocessing Interface is used to download the images and create datasets for an individual category or a dataset of multiple categories.
To detect defects in quartz rods using machine vision technology, we need to get the image through the camera and transmitted it to a computer for next processing. The energy of laser will be produced when it passed t...
详细信息
ISBN:
(纸本)9781538611074
To detect defects in quartz rods using machine vision technology, we need to get the image through the camera and transmitted it to a computer for next processing. The energy of laser will be produced when it passed the object, which makes it difficult to identify the resulting quartz rod boundary. This paper proposes a method of boundary detection based on Hough transform circle, and improves the Hough transform of some algorithms for solving the problem of the large amount of calculation.
暂无评论