Convolutional Neural Network-Based Dimensionality Reduction Method for Image Feature Descriptors Extracted Using Scale-Invariant Feature Transform

2019 ◽  
Vol 56 (14) ◽  
pp. 141008
Author(s):  
周宏浩 Honghao Zhou ◽  
易维宁 Weining Yi ◽  
杜丽丽 Lili Du ◽  
乔延利 Yanli Qiao
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Reduction Method ◽  
Image Feature ◽  
Scale Invariant ◽  
Feature Descriptors ◽  
Invariant Feature ◽  
Feature Transform ◽  
Scale Invariant Feature ◽  
Dimensionality Reduction Method

scholarly journals Dog recognition in public places based on convolutional neural network

2019 ◽  
Vol 15 (5) ◽  
pp. 155014771982967
Author(s):  
Jianquan Ouyang ◽  
Hao He ◽  
Yi He ◽  
Huanrong Tang
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Scale Invariant Feature Transform ◽  
Scale Invariant ◽  
Public Places ◽  
Invariant Feature ◽  
Channel Information ◽  
Feature Transform ◽  
Scale Invariant Feature ◽  
The City

With the increase in the number of dogs in the city, the dogs can be seen everywhere in public places. At the same time, more and more stray dogs appear in public places where dogs are prohibited, which has a certain impact on the city environment and personal safety. In view of this, we propose a novel algorithm that combines dense–scale invariant feature transform and convolutional neural network to solve dog recognition problems in public places. First, the image is divided into several grids; then, the dense–scale invariant feature transform algorithm is used to split and combine the descriptors, and the channel information of the eight directions of the image is extracted as the input of the convolutional neural network; and finally, we design a convolutional neural network based on Adam optimization algorithm and cross-entropy to identify the dog species. The experimental results show that the algorithm can fully combine the advantages of dense–scale invariant feature transform and convolutional neural network to achieve dog recognition in public places, and the correct rate is 94.2%.

scholarly journals Annular Spatial Pyramid Mapping and Feature Fusion-Based Image Coding Representation and Classification

2020 ◽  
Vol 2020 ◽  
pp. 1-9
Author(s):  
Mengxi Xu ◽  
Yingshu Lu ◽  
Xiaobin Wu
Keyword(s):  
Image Classification ◽  
Feature Fusion ◽  
Image Feature ◽  
Classification Model ◽  
Scale Invariant Feature Transform ◽  
Scale Invariant ◽  
Invariant Feature ◽  
Feature Transform ◽  
Scale Invariant Feature ◽  
Spatial Pyramid

Conventional image classification models commonly adopt a single feature vector to represent informative contents. However, a single image feature system can hardly extract the entirety of the information contained in images, and traditional encoding methods have a large loss of feature information. Aiming to solve this problem, this paper proposes a feature fusion-based image classification model. This model combines the principal component analysis (PCA) algorithm, processed scale invariant feature transform (P-SIFT) and color naming (CN) features to generate mutually independent image representation factors. At the encoding stage of the scale-invariant feature transform (SIFT) feature, the bag-of-visual-word model (BOVW) is used for feature reconstruction. Simultaneously, in order to introduce the spatial information to our extracted features, the rotation invariant spatial pyramid mapping method is introduced for the P-SIFT and CN feature division and representation. At the stage of feature fusion, we adopt a support vector machine with two kernels (SVM-2K) algorithm, which divides the training process into two stages and finally learns the knowledge from the corresponding kernel matrix for the classification performance improvement. The experiments show that the proposed method can effectively improve the accuracy of image description and the precision of image classification.

scholarly journals KLASIFIKASI AMERICAN SIGN LANGUAGE MENGGUNAKAN FITUR SCALE INVARIANT FEATURE TRANSFORM DAN JARINGAN SARAF TIRUAN

2020 ◽  
Vol 1 (1) ◽  
pp. 1-11
Author(s):  
Muhammad Restu Alviando ◽  
Muhammad Ezar Al Rivan ◽  
Yoannita Yoannita
Keyword(s):  
Neural Network ◽  
American Sign Language ◽  
Sign Language ◽  
Test Data ◽  
American Sign ◽  
Scale Invariant ◽  
Invariant Feature ◽  
The Neural Network ◽  
Feature Transform ◽  
Scale Invariant Feature

American Sign Language (ASL) is a sign language in the world. This study uses the neural network method as a classification and the scale invariant feature transform (SIFT) as feature extraction. Training data and test data for ASL images were extracted using the SIFT feature, then ANN training was conducted using 17 training functions with 2 hidden layers. There are architecture used [250-5-10-24], [250-5-15-24] and [250-15-15-24] so there are 3 different ANN architectures. Each architecture is performed 3 times so that there are 9 experiments (3 x 3 trials run the program). Determination of the number of neurons concluded by the training function is selected by the best test results on the test data. Based on the training function and the extraction of SIFT features as input values ​​in the neural network it can be concluded that from 17 training functions, trainb with neuron architecture [250-5-10-24] becomes the best training function producing an accuracy value of 95%, precision of 15 % and recall 5%.  

scholarly journals SIFT Image Feature Extraction for an Efficient Image Registration

2020 ◽  
Vol 8 (6) ◽  
pp. 4090-4094
Keyword(s):  
Image Registration ◽  
Image Feature ◽  
Scale Invariant ◽  
Invariant Feature ◽  
Image Registration Algorithm ◽  
Key Points ◽  
Feature Transform ◽  
Scale Invariant Feature ◽  
Ransac Algorithm ◽  
Unit Vectors

This paper presents an image registration algorithm based on SIFT (Scale Invariant Feature Transform).The obtained descriptors and key points by the SIFT confirms that, the algorithm is very robust to scaling, noise, translation and rotation. At the beginning, the key points are extracted from the image. Later to Match the obtained points, dot products between the unit vectors are calculated. Finally, transformation matrix is obtained by applying RANSAC algorithm. Experimental results shows that the algorithm extracts the better key points, which can be used for used for image registration applications.

Sign in / Sign up

Export Citation Format

Share Document