International Journal of Engineering Research and Reviews
ISSN 2348-697X (Online) Vol. 8, Issue 3, pp: (1-10), Month: July - September 2020, Available at: www.researchpublish.com
Local Binary Patterns and Its Extended Variants Hardeep Singh1, Gagandeep2 1
IKG-Punjab Technical University, Jalandhar, Punjab, India
2
Chandigarh Engineering College Landran, Mohali, Punjab, India
Abstract: This paper focuses on the Local Binary Patterns and its various important variants. LBP is a nonparametric descriptor and used to extract, analyze, recognize and classify the different modality images. It summarizes the local patterns of image characteristics efficiently. LBP and its many extended versions have been extensively used in numerous applications of computer vision, image processing, pattern recognition and biomedical field in recent years. Very discriminative and computationally efficient local texture descriptors based on local binary patterns (LBPs) is studied, which led to significant progress in applying texture methods to different problems and applications. The efficiency and usability of the LBP operator and its success in various real world applications has inspired the development of much new powerful LBP variants. In this paper, the important extensions of LBP using local structure of the image are extensively reviewed. Keywords: Local Binary Pattern, Texture, LBP, LTP.
I. INTRODUCTION The texture of images refers to appearance, structure, arrangement of the parts of an object within the image. Images used for diagnostic purposes in clinical field are digital and most probably two dimensional. Texture analysis has been researched since the 1960. In principle it is a technique for evaluating the position and intensity of signal features that means pixels and their gray level intensities. The distribution of these pixels can be computed to produce mathematical parameters which characterize the texture type. The underlying structure of objects is shown in image. These values are also known as texture features. During that time studies showed much important information contained in the distributions of feature values were lost through the usage of these single texture measures. These texture classification methods assume that unknown samples to be classified are always identical to training samples with respect to scale, orientation and gray-scale properties. Real world textures are not like that, those are unpredictably subjected to varying illumination conditions and arbitrary spatial rotations constantly. This showed how unreliable past texture classifications were and their incompetence in handling real world images. Information was first published as part of a comparative study of texture operators in the international conference on pattern recognition [1]. Not to mention, the degree of computational complexity in those algorithms was too high [2]. A very helpful suggestion for future research from then was to develop texture measures which incorporate invariance to real-world factors such as orientation and scale, and can be classified with a low-computational complexity [3].
II. LOCAL BINARY PATTERNS LBP is introduced in 1996 as a comparative study of texture measures with classifications, pattern recognition [4]. LBP is applied in computer vision and image processing. It is used for textual and facial description. It is feature extraction descriptor. This operator acts as an image operator which transforms an image into an array or image of integer labels describing small scale appearance (textures) of image. These label directly or their statistics are used for further analysis. LBP has been found as a powerful feature for texture classification [5]. A. Grayscale image to LBP mask To calculate the LBP descriptor, we convert the input color image to grayscale, since LBP works on grayscale images. For each pixel in the grayscale image, a neighborhood is selected around the current pixel and LBP value is calculated for the pixel using the neighborhood. After calculating the LBP value of the current pixel, we update the corresponding pixel location in the LBP mask (It is of same height and width as the input image) with the LBP value.
Page | 1 Research Publish Journals