Semi-supervised Learning for Hyperspectral Image Classification
半監督學習在高光譜分類中的應用
Student thesis: Doctoral Thesis
Author(s)
Related Research Unit(s)
Detail(s)
Awarding Institution | |
---|---|
Supervisors/Advisors |
|
Award date | 14 Aug 2020 |
Link(s)
Permanent Link | https://scholars.cityu.edu.hk/en/theses/theses(258e7be4-4833-448b-a33b-856de2875773).html |
---|---|
Other link(s) | Links |
Abstract
Hyperspectral imaging collects digital images including hundreds of spectral bands, which often range from the visible to infrared spectrum. The abundant spectral information in hyperspectral images (HSIs) gives rise to promising breakthroughs in multiple application fields, such as military, agriculture, and mineralogy. Over the past decades, many classification methods for hyperspectral images (HSIs) have been proposed, e.g., kernel-based methods, feature extraction-based methods, sparse representation-based methods, etc..
This thesis focuses on considering utilizing predictions based on unlabeled data to help improve the discriminative ability of kernel representation-based classifier learning, sparse representation-based classifier learning and low-rank based feature learning. Specifically, the main methods are summarized as follows.
Firstly, we propose a new kernel-based framework for hyperspectral image classification, namely pseudo-label guided kernel learning (PLKL). The proposed framework is capable of fully utilizing unlabeled samples, making it very effective to handle the task with extremely limited training samples. Specifically, with multiple initial kernels and labeled samples, we first employ SVM classifiers to predict pseudo-labels independently for each unlabeled sample, and consistency voting is applied to the resulting pseudo-labels to select and add a few unlabeled samples to the training set. Then, we refine the kernels to improve their discriminability with the augmented training set and a typical kernel learning method. Such phases are repeated with several iterations until the performance achieves the stable state. Furthermore, we enhance the PLKL in terms of both the computation and memory efficiencies by using a bagging-like strategy, improving its practicality for large-scale datasets. In addition, the proposed framework is quite flexible and general. That is, other advanced kernel-based methods can be incorporated to continuously improve the performance. Experimental results show that the proposed frameworks achieve much higher classification accuracy, compared with state-of-the-art methods. Especially, the classification accuracy improves by more than 5% with very few training samples.
In the second method, we propose a new sparse representation (SR)-based method for hyperspectral image (HSI) classification, namely sparse representation with incremental dictionaries (SRID). Our SRID boosts existing SR-based HSI classification methods significantly, especially when used for the task with extremely limited training samples. Specifically, by exploiting unlabeled pixels with spatial information and multiple-feature-based SR classifiers, we select and add some of them to dictionaries in an iterative manner, such that the representation abilities of the dictionaries are progressively augmented, and likewise more discriminative representations. In addition, to deal with large scale datasets, we use a certainty sampling strategy to control the sizes of dictionaries, such that the computational complexity is well balanced. Experiments over two benchmark datasets show that our proposed method achieves higher classification accuracy than state-of-the-art methods, i.e., the overall classification accuracy can improve by more than 4%.
In the third method, we propose a novel classification scheme for the remotely sensed hyperspectral image (HSI), by comprehensively exploring its unique characteristics, including the local spatial information and low-rankness. The proposed scheme is composed of two modules, i.e., the classification-guided superpixel segmentation and the discriminative low-rank representation, which are iteratively conducted. Specifically, by utilizing the local spatial information and incorporating the predictions from a typical classifier, the first module segments pixels of an input HSI (or its restoration generated by the second module) into superpixels. According to the resulting superpixels, the pixels of the input HSI are then grouped into clusters and fed into our novel discriminative low-rank representation model with an effective numerical solution. Such a model is capable of increasing the intra-class similarity by suppressing the spectral variations locally while promoting the inter-class discriminability globally, leading to a restored HSI with more discriminative pixels. Experimental results on two benchmark datasets demonstrate the significant superiority of the proposed method over state-of-the-art methods, especially for the case with an extremely limited number of training pixels.
This thesis focuses on considering utilizing predictions based on unlabeled data to help improve the discriminative ability of kernel representation-based classifier learning, sparse representation-based classifier learning and low-rank based feature learning. Specifically, the main methods are summarized as follows.
Firstly, we propose a new kernel-based framework for hyperspectral image classification, namely pseudo-label guided kernel learning (PLKL). The proposed framework is capable of fully utilizing unlabeled samples, making it very effective to handle the task with extremely limited training samples. Specifically, with multiple initial kernels and labeled samples, we first employ SVM classifiers to predict pseudo-labels independently for each unlabeled sample, and consistency voting is applied to the resulting pseudo-labels to select and add a few unlabeled samples to the training set. Then, we refine the kernels to improve their discriminability with the augmented training set and a typical kernel learning method. Such phases are repeated with several iterations until the performance achieves the stable state. Furthermore, we enhance the PLKL in terms of both the computation and memory efficiencies by using a bagging-like strategy, improving its practicality for large-scale datasets. In addition, the proposed framework is quite flexible and general. That is, other advanced kernel-based methods can be incorporated to continuously improve the performance. Experimental results show that the proposed frameworks achieve much higher classification accuracy, compared with state-of-the-art methods. Especially, the classification accuracy improves by more than 5% with very few training samples.
In the second method, we propose a new sparse representation (SR)-based method for hyperspectral image (HSI) classification, namely sparse representation with incremental dictionaries (SRID). Our SRID boosts existing SR-based HSI classification methods significantly, especially when used for the task with extremely limited training samples. Specifically, by exploiting unlabeled pixels with spatial information and multiple-feature-based SR classifiers, we select and add some of them to dictionaries in an iterative manner, such that the representation abilities of the dictionaries are progressively augmented, and likewise more discriminative representations. In addition, to deal with large scale datasets, we use a certainty sampling strategy to control the sizes of dictionaries, such that the computational complexity is well balanced. Experiments over two benchmark datasets show that our proposed method achieves higher classification accuracy than state-of-the-art methods, i.e., the overall classification accuracy can improve by more than 4%.
In the third method, we propose a novel classification scheme for the remotely sensed hyperspectral image (HSI), by comprehensively exploring its unique characteristics, including the local spatial information and low-rankness. The proposed scheme is composed of two modules, i.e., the classification-guided superpixel segmentation and the discriminative low-rank representation, which are iteratively conducted. Specifically, by utilizing the local spatial information and incorporating the predictions from a typical classifier, the first module segments pixels of an input HSI (or its restoration generated by the second module) into superpixels. According to the resulting superpixels, the pixels of the input HSI are then grouped into clusters and fed into our novel discriminative low-rank representation model with an effective numerical solution. Such a model is capable of increasing the intra-class similarity by suppressing the spectral variations locally while promoting the inter-class discriminability globally, leading to a restored HSI with more discriminative pixels. Experimental results on two benchmark datasets demonstrate the significant superiority of the proposed method over state-of-the-art methods, especially for the case with an extremely limited number of training pixels.