The Implementation of Parallel Speed up Stereo Matching Based on Reduced Graph Cuts

2014 ◽  
Vol 631-632 ◽  
pp. 486-489
Author(s):  
Ping Zhou ◽  
Mei Liu

The Many stereo matching problems can be converted to energy minimization problem, by establishment of special network graph to obtain the minimum graph cut, and then obtaining the optimal solution. For graph cuts algorithm, complete network graph include all vertices and disparity edges, the computation of time and space is huge. In this paper, we put forward a method by combing local and global stereo matching, set up a reduced network graph by the possible disparities values for each pixel, and then global optimization, to slove the maximum flow polynomial through CUDA parallel computing, greatly reduced the consumption of time and space.

2021 ◽  
Vol 297 ◽  
pp. 01055
Author(s):  
Mohamed El Ansari ◽  
Ilyas El Jaafari ◽  
Lahcen Koutti

This paper proposes a new edge based stereo matching approach for road applications. The new approach consists in matching the edge points extracted from the input stereo images using temporal constraints. At the current frame, we propose to estimate a disparity range for each image line based on the disparity map of its preceding one. The stereo images are divided into multiple parts according to the estimated disparity ranges. The optimal solution of each part is independently approximated via the state-of-the-art energy minimization approach Graph cuts. The disparity search space at each image part is very small compared to the global one, which improves the results and reduces the execution time. Furthermore, as a similarity criterion between corresponding edge points, we propose a new cost function based on the intensity, the gradient magnitude and gradient orientation. The proposed method has been tested on virtual stereo images, and it has been compared to a recently proposed method and the results are satisfactory.


2021 ◽  
Vol 17 (4) ◽  
pp. 1-20
Author(s):  
Serena Wang ◽  
Maya Gupta ◽  
Seungil You

Given a classifier ensemble and a dataset, many examples may be confidently and accurately classified after only a subset of the base models in the ensemble is evaluated. Dynamically deciding to classify early can reduce both mean latency and CPU without harming the accuracy of the original ensemble. To achieve such gains, we propose jointly optimizing the evaluation order of the base models and early-stopping thresholds. Our proposed objective is a combinatorial optimization problem, but we provide a greedy algorithm that achieves a 4-approximation of the optimal solution under certain assumptions, which is also the best achievable polynomial-time approximation bound. Experiments on benchmark and real-world problems show that the proposed Quit When You Can (QWYC) algorithm can speed up average evaluation time by 1.8–2.7 times on even jointly trained ensembles, which are more difficult to speed up than independently or sequentially trained ensembles. QWYC’s joint optimization of ordering and thresholds also performed better in experiments than previous fixed orderings, including gradient boosted trees’ ordering.


Author(s):  
Ruiyang Song ◽  
Kuang Xu

We propose and analyze a temporal concatenation heuristic for solving large-scale finite-horizon Markov decision processes (MDP), which divides the MDP into smaller sub-problems along the time horizon and generates an overall solution by simply concatenating the optimal solutions from these sub-problems. As a “black box” architecture, temporal concatenation works with a wide range of existing MDP algorithms. Our main results characterize the regret of temporal concatenation compared to the optimal solution. We provide upper bounds for general MDP instances, as well as a family of MDP instances in which the upper bounds are shown to be tight. Together, our results demonstrate temporal concatenation's potential of substantial speed-up at the expense of some performance degradation.


2014 ◽  
Vol 592-594 ◽  
pp. 2659-2664 ◽  
Author(s):  
T. Jeyakumar ◽  
R. Gandhinathan

India is the second largest 2W market in the world in terms of sales volumes after China. Motorcycles types that are marketed using sports tag are found to be anchored on performance attributes characterized by visual appeal, higher speeds, heady acceleration and superior ride, handling and braking. The chronograph of the sports segment in the Indian market is plotted. A goal defined design process is developed to produce creative ideas for aesthetic attributes-modern, youthful, aerodynamic, and aggressive. The optimal solution satisfying the aesthetic goal is determined using an operation decision making model based upon weighted generalized mean method. A motorcycle is generally straddled by the rider with manual transmission and can be considered as a constrained workstation. Some ergonomic considerations to fit users of different sizes on the same workstation should be taken into account when designing. A two-dimensional anthropometric data collection approach is followed for riders in India. The obtained anthropometric data concerning riding postures are used for posture analysis using digital human model in CAD software. The mutual trade-off between sporty riding style of the rider and comfort angles have been arrived to set up the final posture of the rider. The detailing of the appearance considering the aesthetic attributes and ergonomics are done. The developed design is aimed at improving appearance and ergonomic performance.


1956 ◽  
Vol 21 ◽  
pp. 156-159
Author(s):  
O. G. S. Crawford

The prudent contributor to a Festschrift will select some subject about which he thinks he knows as much as the professor who is to receive it. That is peculiarly difficult here because of the vast range of Professor Childe's knowledge, both in time and space, far exceeding the present contributor's. This Note is offered as a grateful tribute from one of the many who have been intellectually enriched by his writings and encouraged by his devotion to scholarship. It is little more than an amplification and criticism of the Abbé Breuil's classic Presidential Address to the Prehistoric Society of East Anglia, delivered in 1934; but on the strength of observations made in August and September, 1955, I have come to different conclusions.The Abbé Breuil detected five successive techniques, all of them found on the stones of the Boyne Tombs:(1) Incised thin lines (pl. XIX, B).(2) Picked grooves left rough (pl. XVIII).(3, a) Picked grooves afterwards rubbed smooth; in this and the preceding group ‘it is invariably the line (groove) itself on which the pattern depends, which gives and is the design’.(3, b) Picked areas which ‘only define the limits of the pattern, the surface, left in relief by the cutting down of the background, constituting the actual design’ (pl. xx, B).(4) Rectilinear patterns where also the pattern is residual, consisting of raised ribs, forming triangles or lozenges, left standing by picking away the surrounding surface (pl. xx, A).


1987 ◽  
Vol 116 (1_Suppl) ◽  
pp. S312-S317 ◽  
Author(s):  
G. Benker ◽  
D. Reinwein ◽  
H. Creutzig ◽  
H. Hirche ◽  
W. D. Alexander ◽  
...  

Abstract. In spite of the long-established use of antithyroid drugs, there are many unsettled questions connected with this treatment of Graves' disease. There is a lack of controlled prospective trials studying the results of antithyroid drug therapy while considering the many variables such as disease heterogeneity, regional differences, drug dosage and duration of treatment. Therefore, a multicenter study has been set up in order to compare the effects of two fixed doses of methimazole (10 vs 40 mg) with thyroid hormone supplementation on the clinical, biochemical and immunological course of Graves' disease and on remission rates. Experience accumulated so far suggests that treatment is safe using either 10 or 40 mg of methimazole. While there is a tendency for an advantage of the higher dose within the first weeks (higher effectiveness in controlling hyperthyroidism), this difference is not significant. The impact of dosage on remission rates remains to be shown.


2013 ◽  
Vol 284-287 ◽  
pp. 3428-3432 ◽  
Author(s):  
Yu Hsiu Huang ◽  
Richard Chun Hung Lin ◽  
Ying Chih Lin ◽  
Cheng Yi Lin

Most applications of traditional full-text search, e.g., webpage search, are offline which exploit text search engine to preview the texts and set up related index. However, applications of online realtime full-text search, e.g., network Intrusion detection and prevention systems (IDPS) are too hard to implementation by using commodity hardware. They are expensive and inflexible for more and more occurrences of new virus patterns and the text cannot be previewed and the search must be complete realtime online. Additionally, IDPS needs multi-pattern matching, and then malicious packets can be removed immediately from normal ones without degrading the network performance. Considering the problem of realtime multi-pattern matching, we implement two sequential algorithms, Wu-Manber and Aho-Corasick, respectively over GPU parallel computation platform. Both pattern matching algorithms are quite suitable for the cases with a large amount of patterns. In addition, they are also easier extendable over GPU parallel computation platform to satisfy realtime requirement. Our experimental results show that the throughput of GPU implementation is about five to seven times faster than CPU. Therefore, pattern matching over GPU offers an attractive solution of IDPS to speed up malicious packets detection among the normal traffic by considering the lower cost, easy expansion and better performance.


Sensors ◽  
2018 ◽  
Vol 18 (11) ◽  
pp. 3791
Author(s):  
Tianli Ma ◽  
Song Gao ◽  
Chaobo Chen ◽  
Xiaoru Song

To deal with the problem of multitarget tracking with measurement origin uncertainty, the paper presents a multitarget tracking algorithm based on Adaptive Network Graph Segmentation (ANGS). The multitarget tracking is firstly formulated as an Integer Programming problem for finding the maximum a posterior probability in a cost flow network. Then, a network structure is partitioned using an Adaptive Spectral Clustering algorithm based on the Nyström Method. In order to obtain the global optimal solution, the parallel A* search algorithm is used to process each sub-network. Moreover, the trajectory set is extracted by the Track Mosaic technique and Rauch–Tung–Striebel (RTS) smoother. Finally, the simulation results achieved for different clutter intensity indicate that the proposed algorithm has better tracking accuracy and robustness compared with the A* search algorithm, the successive shortest-path (SSP) algorithm and the shortest path faster (SPFA) algorithm.


2015 ◽  
Author(s):  
Pieraugusto Panzalis ◽  
Andrea Deiana ◽  
Sarah Caronni ◽  
Augusto Navone

Marine Protected Areas (MPAs) are acknowledged globally as effective tools for the protection and management of the marine environment; however, to get effective results it is necessary to set up a proper and continuous mapping of the marine territory, in order to gain detailed knowledge of its different aspects. Therefore, the implementation and maintenance of a modern GIS (Geographic Information System) has become an indispensable task for the MPA of Tavolara - Punta Coda Cavallo to collect, aggregate, classify, and track the conducted mapping activities. Between 2011 and 2012 the sea bottom of the MPA was surveyed using different methods: by means of a multi-beam echo sounder and of a side scan sonar, as well as conducting fast scientific scuba divings with re-breathers and underwater position system technologies. High resolution geodatasets, characterized by a significantly high quality in representing and describing the sea bottom and its habitats, were produced in both feature (scale up to 1:1.250) and raster formats (up to 30cm/pixel for sonar images and 1m/pixel for bathymetry) and they currently constitute the basis of the MPA's GIS, including its 3D applications and its web map services for desktop and mobile devices (iPhone & Android). To update the above described geodatasets during time, acquiring new data on the conservation targets considered in monitoring activities, among which the status of P. oceanica meadows is of the most important ones, a long term mapping plan was realized on the basis of an innovative methodology elaborated by the MPA considering both the wideness of the area and the limited funds available at present. The whole MPA was divided in territorial units by means of a regular grid of square cells having a 100m side with the logic of starting the mapping activities from the mainly important areas and then to spread the surveys up to fill the whole mosaic. All the new data acquired with this methodology could then be mixed, compared and indexed within the same cell and/or in the many already available geodatases, starting from those dated 2006 having a regular grid with square cells of 500m per side.


Sign in / Sign up

Export Citation Format

Share Document