Laws’ mask method has achieved wide acceptance in texture analysis, however it is not robust to noise. Fuzzy filters are well known for denoising applications. This work proposes a noise-robust Laws’ mask descriptor by integrating the exiting fuzzy filters with the traditional Laws’ mask for the improvement of texture classification of noisy texture images. Images are corrupted by adding Gaussian noise of different values. These noisy images are transformed into fuzzy images through fuzzy filters of different windows. Then the texture features are extracted using Laws’ mask descriptor. To investigate the proposed techniques two texture databases i.e. Brodatz and STex are used. The proposals are assessed by comparing the performance of the traditional Laws’ mask descriptor alone and after combined with the fuzzy filters on noisy images. The k-Nearest Neighbor (k-NN) classifier is utilized in the classification task. Results indicate that the proposed approach delivers higher classification accuracy than the traditional Laws’ mask method. Hence, validate that the suggested methods significantly improve the noised texture classification.

Laws的遮罩方法已在纹理分析中获得广泛接受，但它对噪声的鲁棒性不强。模糊滤波器在去噪应用方面是众所周知的。这项工作通过将现有的模糊滤波器与传统的Laws遮罩集成在一起，提出了一种鲁棒的Laws遮罩描述符，以改善嘈杂的纹理图像的纹理分类。通过添加不同值的高斯噪声会损坏图像。这些噪声图像通过不同窗口的模糊过滤器转换为模糊图像。然后使用Laws的蒙版描述符提取纹理特征。为了研究提出的技术，使用了两个纹理数据库，即Brodatz和STex。通过比较单独的传统法律的蒙版描述符的性能以及与模糊滤波器组合后对噪声图像的性能，评估了这些建议。在分类任务中使用了k最近邻（k-NN）分类器。结果表明，与传统的Laws掩膜方法相比，该方法具有更高的分类精度。因此，请验证所建议的方法可以显着改善噪波纹理分类。