In this paper, we address the problem of template matching under affine transformations with general images. Our approach is to search an approximate affine transformation over a binary Galois field. The benefit is that we can avoid matching with huge amount of potential transformations, because they are discretely sampled. However, a Galois field of affine transformation can still be impractical for exhaustive searching. To approach the optimum solution efficiently, we introduce a level-wise adaptive sampling (LAS) method under genetic algorithm framework. In LAS, individuals converge to the global optimum according to a levelwise selection and crossover while the population number is decreased by a population bounding scheme. Specifically, our paper is to infer an approximate affine transformationT from a given candidate set. In the best case,T equals to transformation T . T is the closest transformation to ground truth T among all the candidate transformations. An natural way to estimateT is to minimize SAD. Formally, our purpose can be denoted as:T = arg minAccording to [1], a general affine transformation matrix can be decomposed into T = TrR 2 SR 1 , where R 1 represents matrix operation of 1st rotation, S is scale operation in x-axis and y-axis, R 2 is 2nd rotation, Tr is translation operation in x-axis and y-axis. Transformations over each decomposed DF can be modeled by a Galois field F 2 n , n is a positive integer denoting the length of binary code and 2 n is the field's size. Elements in F 2 n are expressed as binary codes. For clarity, we assume n of each decomposed DF is the same. Each DF's range is then divided into 2 n discrete segments. T ∈ F 2 6n denotes a general affine transformation in 6 DFs. Acceptable margin of error can then be guaranteed on this Galois field. To quantify the error between two transformations T 1 and T 2 , following formula is defined:Level-wise Adaptive Sampling (LAS): LAS aims to achieve a satisfactory error rate instead of testing the complete F 2 n . From the perspective of GA, our problem can be defined as a minimization problem of SAD. In crossover operation of GA, two coded individuals swap certain portions with each other. It is a good method to span search space around a sample point in multiple directions. However, in order to optimizeT in such a broad search space, two major problems should be faced: 1) how to escape from local optimum. 2) how to control the optimization response time.It has been argued in [2] that in order to prevent GA from falling into local optimum, genetic variety should be preserved somehow. Although mutation operation can surely increase the genetic variety randomly, it can also destroy individuals which are potentially to be close toT . In a broad search space, the probability to create a "suitable" diversity is very low and mutation can contrarily slow down the speed of convergence. It is worth noting that in our problem, a large enough number of randomly initialized population keeps sufficient genetic variety for converging toT . Dur...
Evolutionary algorithms (EAs) and swarm algorithms (SAs) have shown their usefulness in solving combinatorial and NP-hard optimization problems in various research fields. However, in the field of computer vision, related surveys have not been updated during the last decade. In this study, inspired by the recent development of deep neural networks in computer vision, which embed large-scale optimization problems, we first describe a literature survey conducted to compensate for the lack of relevant research in this area. Specifically, applications related to the genetic algorithm and differential evolution from EAs, as well as particle swarm optimization and ant colony optimization from SAs and their variants, are mainly considered in this survey.
This paper proposes a decentralized model predictive control method based on a dual decomposition technique. A model predictive control problem for a system with multiple subsystems is formulated as a convex optimization problem. In particular, we deal with the case where the control outputs of the subsystems have coupling constraints represented by linear equalities. A dual decomposition technique is applied to this problem in order to derive the dual problem with decoupled equality constraints. A projected subgradient method is used to solve the dual problem, which leads to a decentralized algorithm. In the algorithm, a small-scale problem is solved at each subsystem, and information exchange is performed in each group consisting of some subsystems. Also, it is shown that the computational complexity in the decentralized algorithm is reduced if the dynamics of the subsystems are all the same.
In this paper, we address the affine template matching of general images. The extensive search space of affine transformations necessitates effective searches of the global optimum. The proposed method utilizes differential evolution (DE), which is a method of metaheuristic optimization, to achieve that goal. Self-adaptive DEs can be useful and are applicable in a wide range of studies as they tune crossover rate and scaling factor (F) themselves over generation iteration. However, this approach is not particularly good for affine template matching because the population often converges to local optima. In order to solve this problem, the population is divided into two equal groups for exploitation and exploration. The former group utilizes current-to-best/1, and the latter group adopts improved current-to-rand/1 for the mutation scheme. Furthermore, the proportion of the population sizes of the two groups are linearly changed on the basis of the best sum of absolute difference error measurements over each generation. These ideas are easy and simple, but experimental results have revealed our method to be more accurate than the state-of-the-art method.
Magnesium hydride is a promising hydrogen source because of its high mass density of hydrogen, 15.2%, when it is hydrolyzed; MgH 2 + 2H 2 O = Mg(OH) 2 + 2H 2 + 277 kJ. However, a magnesium hydroxide, Mg(OH) 2 , layer forms rapidly on the surface of the unreacted MgH 2 as the pH increases, hindering further reaction. The purpose of this study is to find acids that could effectively accelerate the reaction by using a chemical equilibrium analysis where the relationships of pH to concentration of ionized Mg were calculated. For the best performing acid, the calculated and measured relationships were compared, and the effects of acid concentration on hydrogen release were measured. The analysis revealed that citric acid and ethylenediamine-tetraacetic acid were good buffering agents. The calculated and measured relationships between pH and concentration of ionized Mg were in good accord. Hydrogen release improved considerably in a relatively dilute citric acid solution instead of pure distilled water. The maximum amount of hydrogen generated was 1.7 10 3 cm 3 ·g -1 at STP after 30 min. We estimated the exact concentration of citric acid solution for complete MgH 2 hydrolysis by a chemical equilibrium analysis method.
In this paper, we address the problem of projective template matching which aims to estimate parameters of projective transformation. Although homography can be estimated by combining keypoint-based local features and RANSAC, it can hardly be solved with featureless images or high outlier rate images. Estimating the projective transformation remains a difficult problem due to high-dimensionality and strong non-convexity. Our approach is to quantize the parameters of projective transformation with binary finite field and search for an appropriate solution as the final result over the discrete sampling set. The benefit is that we can avoid searching among a huge amount of potential candidates. Furthermore, in order to approximate the global optimum more efficiently, we develop a level-wise adaptive sampling (LAS) method under genetic algorithm framework. With LAS, the individuals are uniformly selected from each fitness level and the elite solution finally converges to the global optimum. In the experiment, we compare our method against the popular projective solution and systematically analyse our method. The result shows that our method can provide convincing performance and holds wider application scope.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.