Commonly, visual inspection tasks in the textile industry are performed by human experts. The major drawback of this type of inspection is the human subjectivity, which affects accuracy and repeatability. Objectivity, accuracy and repeatability can be achieved by analysing visual characteristics of the products using computer vision. Particularly, automatic real time inspection systems based on texture analysis can be implemented using Local Binary Pattern (LBP) techniques. A recent variation of the LBP techniques, named Geometric Local Binary Pattern (GLBP) technique, showed an increase in the performance for detecting small changes of local texture. In this paper a real time implementation of the algorithm is presented by using a Graphic Processing Unit (GPU). The LBP and GLBP techniques are compared in terms of speed and accuracy while implemented on a Central Processing Unit (CPU) and GPU environments. Algorithms are tested for detecting defects in fabrics as well as for evaluating global deviations of texture, which are due to the degradation of the surface in carpets. Results show that higher discriminant power between similar textures is obtained when using the GLBP technique

Video analysis in real time requires fast and efficient algorithms to extract relevant information from a considerable number, commonly 25, of frames per second. Furthermore, robust algorithms for outdoor visual scenes may retrieve correspondent features along the day where a challenge is to deal with lighting changes. Currently, Local Binary Pattern (LBP) techniques are widely used for extracting features due to their robustness to illumination changes and the low requirements for implementation. We propose to compute an automatic threshold based on the distribution of the intensity residuals resulting from the pairwise comparisons when using LBP techniques. The intensity residuals distribution can be modelled by a Generalized Gaussian Distribution (GGD). In this paper we compute the adaptive threshold using the parameters of the GGD. We present a CUDA implementation of our proposed algorithm. We use the LBPSYM technique. Our approach is tested on videos of four different urban scenes with mobilities captured during day and night. The extracted features can be used in a further step to determine patterns, identify objects or detect background. However, further research must be conducted for blurring correction since the scenes at night are commonly blurred due to artificial lighting

This paper compares the speed performance of a set of classic image algorithms for evaluating texture in images by using CUDA programming. We include a summary of the general program mode of CUDA. We select a set of texture algorithms, based on statistical analysis, that allow the use of repetitive functions, such as the Co-ocurrence Matrix, Haralick features and local binary patterns techniques. The memory allocation time between the host and device memory is not taken into account. The results of this approach show a comparison of the texture algorithms in terms of speed when executed on CPU and GPU processors. The comparison shows that the algorithms can be accelerated more than 40 times when implemented using CUDA environment