Method and parallel architecture for extracting the image fractal dimension in real time

1998 ◽  
Author(s):  
Xiangdong Chen ◽  
Wensen Chang ◽  
Zheng Gao
2017 ◽  
Vol 15 (3) ◽  
pp. 657-672 ◽  
Author(s):  
Bingjie Li ◽  
Cunguang Zhang ◽  
Bo Li ◽  
Hongxu Jiang ◽  
Qizhi Xu

2018 ◽  
Vol 103 (3) ◽  
pp. 1941-1963 ◽  
Author(s):  
Deepak Kumar Gupta ◽  
Vijay Kumar Gupta ◽  
Mahesh Chandra ◽  
Gaurav Verma

Sensors ◽  
2018 ◽  
Vol 18 (12) ◽  
pp. 4129 ◽  
Author(s):  
Hassan Mushtaq ◽  
Sajid Gul Khawaja ◽  
Muhammad Usman Akram ◽  
Amanullah Yasin ◽  
Muhammad Muzammal ◽  
...  

Clustering is the most common method for organizing unlabeled data into its natural groups (called clusters), based on similarity (in some sense or another) among data objects. The Partitioning Around Medoids (PAM) algorithm belongs to the partitioning-based methods of clustering widely used for objects categorization, image analysis, bioinformatics and data compression, but due to its high time complexity, the PAM algorithm cannot be used with large datasets or in any embedded or real-time application. In this work, we propose a simple and scalable parallel architecture for the PAM algorithm to reduce its running time. This architecture can easily be implemented either on a multi-core processor system to deal with big data or on a reconfigurable hardware platform, such as FPGA and MPSoCs, which makes it suitable for real-time clustering applications. Our proposed model partitions data equally among multiple processing cores. Each core executes the same sequence of tasks simultaneously on its respective data subset and shares intermediate results with other cores to produce results. Experiments show that the computational complexity of the PAM algorithm is reduced exponentially as we increase the number of cores working in parallel. It is also observed that the speedup graph of our proposed model becomes more linear with the increase in number of data points and as the clusters become more uniform. The results also demonstrate that the proposed architecture produces the same results as the actual PAM algorithm, but with reduced computational complexity.


2002 ◽  
Vol 199 ◽  
pp. 506-507
Author(s):  
Carlo Rosolen ◽  
Alain Lecacheux ◽  
Eric Gerard ◽  
Vincent Clerc ◽  
Laurent Denis

Radio astronomy in the decameter to centimeter wavelength range is facing new challenges because of man made interferences due to increasing needs in telecommunications. At the Radioastronomy department of Paris Meudon Observatory, we have been working since four years on high dynamic range digital receivers based on Digital Signal Processors (DSP). The first achievement is a digital spectro- polarimeter devoted to spectroscopy of astrophysical radiation in decameter range, now in operation at the Nancay Decameter array. The block diagram of the receiver includes a high dynamic range analogue section followed by a 12 bits analogue to digital converter. The digital part makes use of high power, programmable digital circuits for signal processing, arranged in a dedicated parallel architecture, able to compute in real time the power spectrum and the correlation of the input signals. This receiver was also used, as spectrometer backend, at Nancay decimetric radiotelescope and has performed very well in the presence of very strong interferences. We are presently working on a new digital receiver with broader bandwidth. The objective is 2 × 25 MHz band with at least 60 dB dynamic range. This new receiver will use additional computation power in order to recognise and avoid man made interferences which corrupt the radio astronomical signal. At the Nancay Radioastronomy Observatory, we have started to develop a new digital configurable receiver with 8 times 25 MHz band and ten thousand channels. For low frequency radioastronomy, direct spectrum computation technique is really powerful and offers new capabilities for real time interferences excision. Fig. 1 shows pulsar observations in the presence of interference made with the DSP receiver on the UTR-2 radiotelescope. Fig. 2 shows the effect of satellite interfernce on OH observations made with the Nancay telescope. Fig. 3 shows the block diagram of the DSP system and demonstrates how offline excision of interference in the frequency time-domain enables recovery of the signal. The final spectrum had 960 minutes integration on and off source and took 8045 minutes of procession on a 450 MHz Pentium II.


2014 ◽  
Vol 2014 ◽  
pp. 1-13
Author(s):  
Jinwei Wang ◽  
Xirong Ma ◽  
Yuanping Zhu ◽  
Jizhou Sun

The active appearance model (AAM) is one of the most powerful model-based object detecting and tracking methods which has been widely used in various situations. However, the high-dimensional texture representation causes very time-consuming computations, which makes the AAM difficult to apply to real-time systems. The emergence of modern graphics processing units (GPUs) that feature a many-core, fine-grained parallel architecture provides new and promising solutions to overcome the computational challenge. In this paper, we propose an efficient parallel implementation of the AAM fitting algorithm on GPUs. Our design idea is fine grain parallelism in which we distribute the texture data of the AAM, in pixels, to thousands of parallel GPU threads for processing, which makes the algorithm fit better into the GPU architecture. We implement our algorithm using the compute unified device architecture (CUDA) on the Nvidia’s GTX 650 GPU, which has the latest Kepler architecture. To compare the performance of our algorithm with different data sizes, we built sixteen face AAM models of different dimensional textures. The experiment results show that our parallel AAM fitting algorithm can achieve real-time performance for videos even on very high-dimensional textures.


2010 ◽  
Vol 02 (04) ◽  
pp. 509-520 ◽  
Author(s):  
SY-SANG LIAW ◽  
FENG-YUAN CHIU

Real nonstationary time sequences are in general not monofractals. That is, they cannot be characterized by a single value of fractal dimension. It has been shown that many real-time sequences are crossover-fractals: sequences with two fractal dimensions — one for the short and the other for long ranges. Here, we use the empirical mode decomposition (EMD) to decompose monofractals into several intrinsic mode functions (IMFs) and then use partial sums of the IMFs decomposed from two monofractals to construct crossover-fractals. The scale-dependent fractal dimensions of these crossover-fractals are checked by the inverse random midpoint displacement method (IRMD).


Sign in / Sign up

Export Citation Format

Share Document