Local Derivative Vector Pattern: Hybrid Pattern for Content-Based Medical Image Retrieval

This paper examines a hybrid pattern i.e. Local derivative Vector pattern and comparasion of this pattern over other different patterns for content-based medical image retrieval. In recent years Pattern-based texture analysis has significant popularity for a variety of tasks like image recognition, image and texture classification, and object detection, etc. In literature, different patterns exist for texture analysis. This paper aims at forming a hybrid pattern compared in terms of precision, recall and F1-score with different patterns like Local Binary Pattern (LBP), Local Derivative Pattern (LDP), Completed Local Binary Pattern (CLBP), Local Tetra Pattern (LTrP), Local Vector Pattern (LVP) and Local Anisotropic Pattern (LAP) which were applied on medical images for image retrieval. The proposed method is evaluated on different modalities of medical images. The results of the proposed hybrid pattern show biased performance compared to the state-of-the-art. So this can further extended with other pattern to form a hybrid pattern.

Real-Time Implementation and Performance Optimization of Local Derivative Pattern Algorithm on GPUs

<p><span>Pattern based texture descriptors are widely used in Content Based Image Retrieval (CBIR) for efficient retrieval of matching images. Local Derivative Pattern (LDP), a higher order local pattern operator, originally proposed for face recognition, encodes the distinctive spatial relationships contained in a local region of an image as the feature vector. LDP efficiently extracts finer details and provides efficient retrieval however, it was proposed for images of limited resolution. Over the period of time the development in the digital image sensors had paid way for capturing images at a very high resolution. LDP algorithm though very efficient in content-based image retrieval did not scale well when capturing features from such high-resolution images as it becomes computationally very expensive. This paper proposes how to efficiently extract parallelism from the LDP algorithm and strategies for optimally implementing it by exploiting some inherent General-Purpose Graphics Processing Unit (GPGPU) characteristics. By optimally configuring the GPGPU kernels, image retrieval was performed at a much faster rate. The LDP algorithm was ported on to Compute Unified Device Architecture (CUDA) supported GPGPU and a maximum speed up of around 240x was achieved as compared to its sequential counterpart.</span></p>

Deteksi Serangan Spoofing Pada Citra Wajah menggunakan Ekstraksi Ciri Local Derivative Pattern

Pada penelitian ini, diusulkan sistem pendeteksi serangan spoofing pada citra wajah manusia menggunakan metode ekstraksi ciri Local Derivative Pattern (LDP). Metode klasifikasi yang digunakan adalah k-Nearest Neighbour (k-NN) dan Support Vector Machine (SVM). Penelitian ini menggunakan NUAA Imposter and Photograph Database sebagai datasetnya. Parameter optimal untuk ekstraksi ciri menggunakan LDP, adalah sebagai berikut: LDP orde ke-2 dengan radius bernilai 5 yang bersifat overlapping non-uniform menggunakan algoritma klasifikasi SVM dengan kernel Radial Basis Function. Performansi terbaik didapatkan menggunakan F1-Score sebesar 99.8%. Pola uniform pada LDP mempercepat waktu komputasi dengan rata-rata 2.09 detik, sedangkan waktu komputasi pola non-uniform yaitu 5.49 detik.

Texture-Based Land Cover Classification Algorithm Using Hidden Markov Model for Multispectral Data

In this chapter, the concept of stochastic optimal control is well explored using hidden Markov model (HMM) in classifying land covers of remotely sensed images. The features of land covers can be colour, shape, and texture. Texture is a useful feature in land cover classification. A texture-based land cover classification algorithm using HMM has been proposed. The local derivative pattern (LDP) texture descriptor for gray level images has been extended as multivariate local derivative pattern (MLDP) for remotely sensed images in this chapter. Experiments were conducted on IRS P6 LISS-IV data and the results were evaluated based on the classification accuracy and compared against the three existing methods such as wavelet, MLDP and colour gray level co-occurrence matrix (CGLCM). The results indicate that the proposed algorithm achieves a classification accuracy of 88.75%.