Self-adaptive hybrid PSO-GA method for change detection under varying contrast conditions in satellite images

Abstract

Abstract-This paper proposes a new unsupervised satellite change detection method, which is robust to illumination changes. To achieve this, firstly, a preprocessing strategy is used to remove illumination artifacts and results in less false detection than traditional threshold-based algorithms. Then, we use the corrected input data to define a new fitness function based on the difference image. The purpose of using Self-Adaptive Hybrid Particle Swarm Optimization-Genetic Algorithm (SAPSOGA) is to combine two meta-heuristic optimization algorithms to search and find the feasible solution in the NP-hard change detection problem rapidly and efficiently. The hybrid algorithm is employed by letting the GA and PSO run simultaneously and similarities of GA and PSO have been considered to implement the algorithm, i.e. the population. In the SAPSOGA employed, in each iteration/generation the two population based algorithms share different amount of information or individual(s) between themselves. Thus, each algorithm informs each other about their best optimum results (fitness values and solution representations) which are obtained in their own population. The fitness function is minimized by using binary based SAPSOGA approach to produce binary change detection masks in each iteration to obtain the optimal change detection mask between two multi temporal multi spectral landsat images. The proposed approach effectively optimizes the change detection problem and finds the final change detection mask.

Change Detection in Multispectral Landsat Images Using Multiobjective Evolutionary Algorithm

“…The proposed method is compared with EM-based [21], GA-based [5], ERGAS-based [8], and Particle Swarm Optimization (PSO)-GA-based [6] change detection methods. The proposed method, PSO-GA-based, and ERGAS-based methods use the images in RGB color space whereas the Expectation Maximization (EM)-based and GA-based methods use gray-scale images.…”

“…The EM-based and ERGAS-based methods do not have any parameter to set. In other two methods, we use the parameters given in [6]. In the proposed approach, the NSGA-II is used with the population size of 30, the crossover rate of 0.8, the mutation rate of 0.01, and the iteration number of 25 000.…”

“…These parameters are selected empirically. Besides the qualitative tests, the change detection accuracies are evaluated using some standardized measures, including false alarm rate (P FA ), missed detection rate (P MD ), and total error (P TE ) [6].…”

“…In general, change detection methods can be classified into three main categories: supervised [3], semisupervised [4], and unsupervised [5], [6] approaches. In the supervised and semisupervised approaches, a training set with accurate ground truth must be used to calculate the transformation function via a classifier, which will be used to flag each pixel as changed or unchanged.…”

“…In the first step, the similarity between images is computed using an appropriate mathematical model, such as: image differencing, log-ratio, change vector analysis (CVA), correlation coefficient (CC), erreur relative globale adimensionnelle de synthese (ERGAS), and so on. The second step is used to obtain binary change mask, which can be achieved by optimization methods [5], [6], thresholding [8], clustering approach [9], and many others [10]. For instance, Ghosh et al [11] propose an unsupervised change detection technique for multispectral remote sensing images.…”