Yifeng Niu

A Novel Approach to Image Fusion Based on Multi-Objective Optimization

Most approaches to image fusion determine the building of image fusion model based on experience, and the parameter configuration of the fusion model is somewhat arbitrary. In this paper, a novel approach to image fusion based on multi-objective optimization was presented, which could achieve the optimal fusion indices through optimizing the fusion parameters. First the uniform model of image fusion in DWT (discrete wavelet transform) domain was established; then the proper evaluation indices of image fusion were given; and finally the adaptive multi-objective particle swarm optimization (AMOPSO) was introduced to search the optimal fusion parameters. Experiment results show that AMOPSO has a higher convergence speed and better exploratory capabilities than MOPSO, and that the approach to image fusion based on AMOPSO realizes the Pareto optimal image fusion

Airborne Infrared and Visible Image Fusion for Target Perception Based on Target Region Segmentation and Discrete Wavelet Transform

Infrared and visible image fusion is an important precondition of realizing target perception for unmanned aerial vehicles (UAVs), then UAV can perform various given missions. Information of texture and color in visible images are abundant, while target information in infrared images is more outstanding. The conventional fusion methods are mostly based on region segmentation; as a result, the fused image for target recognition could not be actually acquired. In this paper, a novel fusion method of airborne infrared and visible image based on target region segmentation and discrete wavelet transform (DWT) is proposed, which can gain more target information and preserve more background information. The fusion experiments are done on condition that the target is unmoving and observable both in visible and infrared images, targets are moving and observable both in visible and infrared images, and the target is observable only in an infrared image. Experimental results show that the proposed method can generate better fused image for airborne target perception.

An adaptive multi-objective particle swarm optimization for color image fusion

A novel algorithm of adaptive multi-objective particle swarm optimization (AMOPSO-II) is proposed and used to search the optimal color image fusion parameters, which can achieve the optimal fusion indices. First the algorithm of AMOPSO-II is designed; then the model of color image fusion in YUV color space is established, and the proper evaluation indices are given; and finally AMOPSO-II is used to search the optimal fusion parameters. AMOPSO-II uses a new crowding operator to improve the distribution of nondominated solutions along the Pareto front, and uses the uniform design to obtain the optimal combination of the parameters of AMOPSO-II. Experimental results indicate that AMOPSO-II has better exploratory capabilities than MOPSO and AMOPSO-I, and that the approach to color image fusion based on AMOPSO-II realizes the Pareto optimal color image fusion.

Multi-Objective Wavelet-Based Pixel-Level Image Fusion Using Multi-Objective Constriction Particle Swarm Optimization

In most methods of pixel-level image fusion, determining how to build the fusion model is usually based on people’s experience, and the configuration of fusion parameters is somewhat arbitrary. In this chapter, a novel method of multi-objective pixel-level image fusion is presented, which can overcome the limitations of conventional methods, simplify the fusion model, and achieve the optimal fusion metrics. First the uniform model of pixel-level image fusion based on discrete wavelet transform is established, two fusion rules are designed; then the proper evaluation metrics of pixel-level image fusion are given, new conditional mutual information is proposed, which can avoid the information overloaded; finally the fusion parameters are selected as the decision variables and the multi-objective constriction particle swarm optimization (MOCPSO) is proposed and used to search the optimal fusion parameters. MOCPSO not only uses mutation operator to avoid earlier convergence, but also uses a new crowding operator to improve the distribution of nondominated solutions along the Pareto front, and introduces the uniform design to obtain the optimal parameter combination. The experiments of MOCPSO test, multi-focus image fusion, blind image fusion, multi-resolution image fusion, and color image fusion are conducted. Experimental results indicate that MOCPSO has a higher convergence speed and better exploratory capabilities than MOPSO, especially when the number of objectives is large, and that the fusion method based on MOCPSO is is suitable for many types of pixel-level image fusion and can realize the Pareto optimal image fusion.

Optimizing the number of decomposition levels for wavelet-based multifocus image fusion

The performance of wavelet-based multifocus image fusion depends on the number of decomposition levels in the wavelet transform. Too few decomposition levels result in poor spatial quality in the fused images. On the other hand, too many levels induce distortion between the original and the fused images. In general, fusion of images with larger resolution requires a higher number of decomposition levels. In this paper, a multiobjective optimization approach to determine the optimal number of decomposition levels is presented. First the optimization flow of decomposition levels for wavelet-based multifocus image fusion is presented; then the proper evaluation metrics for multifocus image fusion are given, new conditional mutual information is proposed, which can avoid the information overlapping; and finally the algorithm of multiobjective particle swarm optimization using the local best model is designed to search the optimal number of decomposition levels. In the experiments it is found that the optimal decomposition level is not a fixed value, but rather, changes with the characteristics of the original images. The experimental results also show that using the method of multiobjective optimization can effectively obtain the optimal number of decomposition levels.

Multiobjective Constriction Particle Swarm Optimization and Its Performance Evaluation

A novel multiobjective constriction particle swarm optimization (MOCPSO) is presented. MOCPSO not only uses mutation operator to avoid earlier convergence and uses adaptive weight to raise the search capacity, but also uses a new crowding operator to improve the distribution of nondominated solutions along the Pareto front, and uses the uniform design to obtain the optimal parameter combination. The sound evaluation criteria for multiobjective optimization algorithm are given, and some typical test functions are introduced. Experimental results show that MOCPSO has faster convergent speed and better search capacity than other multiobjective particle swarm optimization algorithms, especially when there are more than two objectives.

Modeling and characterizing of unmanned aerial vehicles autonomy

How to describe and measure unmanned aerial vehicle (UAV) autonomy is of primary importance in the field of autonomous control of UAVs. One of the most challenging obstacles lies in the reasonable description framework of UAV autonomy. In this paper, the existing typical classification models of autonomous control levels are presented and compared firstly. With the construction of task-level model and decision-process model of UAV autonomous control system, a novel classification framework based on the two models is proposed. Under this framework, task-level model should reflect UAV autonomy as perceived by an external observer, while decision-process model considers more internal decision-making aspects. According to the description framework, a detailed autonomous control levels with two dimensions is proposed. Finally, some future directions and difficulties of the issue about evaluating UAV autonomy are also addressed.

Multi-resolution Image Fusion Using AMOPSO-II

Most approaches to multi-resolution image fusion are based on experience, and the fusion results are not the optimal. In this paper, a new approach to multi-resolution image fusion based on AMOPSO-II (Adaptive Multi-Objective Particle Swarm Optimization) is presented, which can achieve the optimal fusion results through optimizing the fusion parameters. First the uniform model of multi-resolution image fusion in DWT (Discrete Wavelet Transform) domain is established; then the proper evaluation indices of multiresolution image fusion are given; and finally AMOPSO-II is proposed and used to search the fusion parameters. AMOPSO-II not only uses an adaptive mutation operator and an adaptive inertia weight to raise the search capacity, but also uses a new crowding operator to improve the distribution of nondominated solutions along the Pareto front, and uses the uniform design to obtain the optimal combination of the parameters of AMOPSO-II. Results show that AMOPSO-II has better exploratory capabilities than AMOPSO-I, and that the approach to multi-resolution image fusion based on AMOPSO-II realizes the Pareto optimal multi-resolution image fusion.

Distributed Cooperative Planning for multiple UAVs Based on Agent Negotiation

Aiming to overcome the limitations of centralized planning methods, distributed planning approaches based on agent negotiation are proposed for a team of unmanned aerial vehicles (UAVs) to execute tasks cooperatively. To solve the problem, a two-stage approach is presented. Firstly, contract net protocol is applied, so that UAVs plan independently and negotiate mutually to realize dynamic task allocation. By integrating several types of contracts to work together, task allocation in complex battlefield environment can be solved more efficiently. Secondly, based on the detection of the relations among tasks, a novel compensatory negotiation mechanism is presented to coordinate the task execution process of UAVs whose tasks are correlative. Thus, UAVs can achieve their tasks more effectively with the assistance of others. Finally, simulation results demonstrate the validity of the approach proposed

Contextual Hierarchical Part-Driven Conditional Random Field Model for Object Category Detection

Even though several promising approaches have been proposed in the literature, generic category-level object detection is still challenging due to high intraclass variability and ambiguity in the appearance among different object instances. From the view of constructing object models, the balance between flexibility and discrimination must be taken into consideration. Motivated by these demands, we propose a novel contextual hierarchical part-driven conditional random field (CRF) model, which is based on not only individual object part appearance but also model contextual interactions of the parts simultaneously. By using a latent two-layer hierarchical formulation of labels and a weighted neighborhood structure, the model can effectively encode the dependencies among object parts. Meanwhile, beta-stable local features are introduced as observed data to ensure the discriminative and robustness of part description. The object category detection problem can be solved in a probabilistic framework using a supervised learning method based on maximum a posteriori (MAP) estimation. The benefits of the proposed model are demonstrated on the standard dataset and satellite images.