Two novel local binary pattern descriptors for texture analysis

Texture classification for visual data using transfer learning

The texture is the most fundamental aspect of a picture that contributes to its recognition. Computer vision challenges such as picture identification and segmentation are built on the foundation of texture analysis. Various images of satellite, forestry, medical etc. have been identifiable because of textures. This work aims to offer texture classification models that will outperform previously presented methods. In this work, transfer learning was applied to attain this goal. MobileNetV3 and InceptionV3 are the two pre-trained models employed. Brodatz, Kylberg, and Outex texture datasets were used to evaluate the models. The models achieved excellent results and achieved the objective in most cases. Classification accuracy obtained for the Kylberg dataset were 100% and 99.89%. For the Brodatz dataset, the classification accuracy obtained was 99.83% and 99.94%. For the Outex datasets, the classification accuracy obtained was 99.48% and 99.48%. The model outputs the corresponding label of the texture of the image.

On the Quantification of Visual Texture Complexity

Complexity is one of the major attributes of the visual perception of texture. However, very little is known about how humans visually interpret texture complexity. A psychophysical experiment was conducted to visually quantify the seven texture attributes of a series of textile fabrics: complexity, color variation, randomness, strongness, regularity, repetitiveness, and homogeneity. It was found that the observers could discriminate between the textures with low and high complexity using some high-level visual cues such as randomness, color variation, strongness, etc. The results of principal component analysis (PCA) on the visual scores of the above attributes suggest that complexity and homogeneity could be essentially the underlying attributes of the same visual texture dimension, with complexity at the negative extreme and homogeneity at the positive extreme of this dimension. We chose to call this dimension visual texture complexity. Several texture measures including the first-order image statistics, co-occurrence matrix, local binary pattern, and Gabor features were computed for images of the textiles in sRGB, and four luminance-chrominance color spaces (i.e., HSV, YC_bC_r, Ohta’s I₁I₂I₃, and CIELAB). The relationships between the visually quantified texture complexity of the textiles and the corresponding texture measures of the images were investigated. Analyzing the relationships showed that simple standard deviation of the image luminance channel had a strong correlation with the corresponding visual ratings of texture complexity in all five color spaces. Standard deviation of the energy of the image after convolving with an appropriate Gabor filter and entropy of the co-occurrence matrix, both computed for the image luminance channel, also showed high correlations with the visual data. In this comparison, sRGB, YC_bC_r, and HSV always outperformed the I₁I₂I₃ and CIELAB color spaces. The highest correlations between the visual data and the corresponding image texture features in the luminance-chrominance color spaces were always obtained for the luminance channel of the images, and one of the two chrominance channels always performed better than the other. This result indicates that the arrangement of the image texture elements that impacts the observer’s perception of visual texture complexity cannot be represented properly by the chrominance channels. This must be carefully considered when choosing an image channel to quantify the visual texture complexity. Additionally, the good performance of the luminance channel in the five studied color spaces proves that variations in the luminance of the texture, or as one could call the luminance contrast, plays a crucial role in creating visual texture complexity.

Brain tumor classification using modified local binary patterns (LBP) feature extraction methods

Automatic classification of brain tumor types is very important for accelerating the treatment process, planning and increasing the patient's survival rate. Today, MR images are used to determine the type of brain tumor. Manual diagnosis of brain tumor type depends on the experience and sensitivity of radiologists. Therefore, researchers have developed many brain tumor classification models to minimize the human factor. In this study, two different feature extraction (nLBP and αLBP) approaches were used to classify the most common brain tumor types; Glioma, Meningioma, and Pituitary brain tumors. nLBP is formed based on the relationship for each pixel around the neighbors. The nLBP method has a d parameter that specifies the distance between consecutive neighbors for comparison. Different patterns are obtained for different d parameter values. The αLBP operator calculates the value of each pixel based on an angle value. The angle values used for calculation are 0, 45, 90 and 135. To test the proposed methods, it was applied to images obtained from the brain tumor database collected from Nanfang Hospital, Guangzhou, China, and Tianjin Medical University General Hospital between the years of 2005 and 2010. The classification process was performed by using K-Nearest Neighbor (Knn) and Artificial Neural Networks (ANN), Random Forest (RF), A1DE, Linear Discriminant Analysis (LDA) classification methods, with the feature matrices obtained with nLBP, αLBP and classical LBP from the images in the data set. The highest success rate in brain tumor classification was 95.56% with the nLBPd = 1 feature extraction method and Knn model.