Evaluation of noise robustness for local binary pattern descriptors in texture classification

Abstract

Local binary pattern (LBP) operators have become commonly used texture descriptors in recent years. Several new LBP-based descriptors have been proposed, of which some aim at improving robustness to noise. To do this, the thresholding and encoding schemes used in the descriptors are modified. In this article, the robustness to noise for the eight following LBP-based descriptors are evaluated; improved LBP, median binary patterns (MBP), local ternary patterns (LTP), improved LTP (ILTP), local quinary patterns, robust LBP, and fuzzy LBP (FLBP). To put their performance into perspective they are compared to three well-known reference descriptors; the classic LBP, Gabor filter banks (GF), and standard descriptors derived from gray-level co-occurrence matrices. In addition, a roughly five times faster implementation of the FLBP descriptor is presented, and a new descriptor which we call shift LBP is introduced as an even faster approximation to the FLBP. The texture descriptors are compared and evaluated on six texture datasets; Brodatz, KTH-TIPS2b, Kylberg, Mondial Marmi, UIUC, and a Virus texture dataset. After optimizing all parameters for each dataset the descriptors are evaluated under increasing levels of additive Gaussian white noise. The discriminating power of the texture descriptors is assessed using tenfolded cross-validation of a nearest neighbor classifier. The results show that several of the descriptors perform well at low levels of noise while they all suffer, to different degrees, from higher levels of introduced noise. In our tests, ILTP and FLBP show an overall good performance on several datasets. The GF are often very noise robust compared to the LBP-family under moderate to high levels of noise but not necessarily the best descriptor under low levels of added noise. In our tests, MBP is neither a good texture descriptor nor stable to noise.

Using filter banks in Convolutional Neural Networks for texture classification

“…We replicate the setup of [14]. We randomly choose one of 12 available orientations for each image and split the images into four subimages which results in 28x160x4 = 17920 images of size 288x288 which we resize to 256x256.…”

“…ImNet-S2 [14] 99.8 ±0.1 [3] 75.5 ±0.8 [4] 81.5 ±2.0 [4] ---containing object-like images. This is due to the fact that the first convolution layer extracts extremely simple features (mainly edges) and acts like a Gabor filter which is not robust at classifying object datasets.…”

Repeating average filter for noisy texture classification

“…As it is mentioned in some papers [55], the performance of Gabor lter and co-occurrence matrix methods signi cantly declines for noisy textures. Table 7 shows some of these results.…”

“…Some methods such as GLCM that use cooccurrence matrix are not rotation-invariant and they are sensitive to noise. Therefore, if they are used for these textures, the classi cation accuracy decreases signi cantly [55].…”

“…These methods are some of the most popular LBPs that are resistant to noise. Kylberg et al reviewed and compared most of them in [55].…”