当前位置: X-MOL 学术J. Math. Imaging Vis. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Radon Cumulative Distribution Transform Subspace Modeling for Image Classification
Journal of Mathematical Imaging and Vision ( IF 1.3 ) Pub Date : 2021-08-05 , DOI: 10.1007/s10851-021-01052-0
Mohammad Shifat-E-Rabbi 1 , Xuwang Yin 2 , Abu Hasnat Mohammad Rubaiyat 2 , Shiying Li 1 , Soheil Kolouri 3 , Akram Aldroubi 4 , Jonathan M Nichols 5 , Gustavo K Rohde 6
Affiliation  

We present a new supervised image classification method applicable to a broad class of image deformation models. The method makes use of the previously described Radon Cumulative Distribution Transform (R-CDT) for image data, whose mathematical properties are exploited to express the image data in a form that is more suitable for machine learning. While certain operations such as translation, scaling, and higher-order transformations are challenging to model in native image space, we show the R-CDT can capture some of these variations and thus render the associated image classification problems easier to solve. The method—utilizing a nearest-subspace algorithm in the R-CDT space—is simple to implement, non-iterative, has no hyper-parameters to tune, is computationally efficient, label efficient, and provides competitive accuracies to state-of-the-art neural networks for many types of classification problems. In addition to the test accuracy performances, we show improvements (with respect to neural network-based methods) in terms of computational efficiency (it can be implemented without the use of GPUs), number of training samples needed for training, as well as out-of-distribution generalization. The Python code for reproducing our results is available at Shifat-E-Rabbi et al. (Python code implementing the Radon cumulative distribution transform subspace model for image classification. https://github.com/rohdelab/rcdt_ns_classifier).



中文翻译:

用于图像分类的氡累积分布变换子空间建模

我们提出了一种新的监督图像分类方法,适用于广泛的图像变形模型。该方法利用之前描述的对图像数据的氡累积分布变换 (R-CDT),利用其数学特性以更适合机器学习的形式表达图像数据。虽然某些操作(例如平移、缩放和高阶变换)在原生图像空间中建模具有挑战性,但我们表明 R-CDT 可以捕获其中的一些变化,从而使相关的图像分类问题更容易解决。该方法在 R-CDT 空间中使用最近子空间算法,实现简单、非迭代、无需调整超参数、计算效率高、标签效率高、并为许多类型的分类问题提供与最先进的神经网络具有竞争力的准确性。除了测试精度性能外,我们还展示了计算效率(可以在不使用 GPU 的情况下实现)、训练所需的训练样本数量以及输出方面的改进(相对于基于神经网络的方法)分布泛化。Shifat-E-Rabbi 等人提供了用于重现我们结果的 Python 代码。(实现用于图像分类的氡累积分布变换子空间模型的 Python 代码。https://github.com/rohdelab/rcdt_ns_classifier)。我们在计算效率(可以在不使用 GPU 的情况下实现)、训练所需的训练样本数量以及分布外泛化方面展示了改进(相对于基于神经网络的方法)。Shifat-E-Rabbi 等人提供了用于重现我们结果的 Python 代码。(实现用于图像分类的氡累积分布变换子空间模型的 Python 代码。https://github.com/rohdelab/rcdt_ns_classifier)。我们在计算效率(可以在不使用 GPU 的情况下实现)、训练所需的训练样本数量以及分布外泛化方面展示了改进(相对于基于神经网络的方法)。Shifat-E-Rabbi 等人提供了用于重现我们结果的 Python 代码。(实现用于图像分类的氡累积分布变换子空间模型的 Python 代码。https://github.com/rohdelab/rcdt_ns_classifier)。

更新日期:2021-08-09
down
wechat
bug