Ordinary object recognizing method based on 2D and 3D SIFT feature fusion

A feature fusion, object recognition technology, applied in character and pattern recognition, instruments, computer parts and other directions

Active Publication Date: 2015-06-17
SOUTHEAST UNIV
View PDF3 Cites 61 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a general object recognition method based on 2D and 3D SIFT feature fusion, combining two-dimensional and three-dimensional features to fuse various object information, which can effectively reduce the The problem of low recognition rate of feature recognition algorithm, in the case of high similarity between classes and small differences within classes, it still has a high recognition accuracy rate

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Ordinary object recognizing method based on 2D and 3D SIFT feature fusion
  • Ordinary object recognizing method based on 2D and 3D SIFT feature fusion
  • Ordinary object recognizing method based on 2D and 3D SIFT feature fusion

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0077] Embodiment 1 Recognition Algorithm Framework

[0078] A general object recognition method based on fusion of 2D and 3D SIFT features mainly includes the following steps:

[0079] 1) Feature extraction and representation:

[0080] Feature extraction and representation are the basis of object recognition. How to extract stable and effective features is the focus and difficulty in feature extraction research. Only by selecting good features can the best recognition results be obtained under the condition of limited training samples. Generally, there are a large number of objects, and it is impossible to build a model library for each object. At the same time, the shape and color of each object of each class are also very different, so the extracted object features must meet the following conditions: 1) Make the difference between the classes the largest, that is, the most It can represent the characteristics of each type of object that are different from other types of ob...

Embodiment 2

[0148] Embodiment 2 algorithm flow

[0149] like figure 2 Shown is a schematic flow chart of a general object recognition algorithm based on 2D and 3D SIFT feature fusion. The general object recognition process proposed by the present invention mainly includes two stages of offline training and online recognition. The following is a detailed description of the training link and the recognition link in the flow chart.

[0150] 1. Training algorithm process:

[0151] 1.1 Offline training phase:

[0152] 1.1.1 After the offline training starts, the image p corresponding to the i-th class object in the object image database i The point cloud pc corresponding to the i-th object in the object point cloud library i , i=1, 2, K n, n represents the number of training sample categories, first extract the 2D and 3D SIFT features corresponding to n types of training samples, recorded as F_R={f i _R,i=1,2,K n},R∈(2D,3D), where, f i _2D is m i *128 feature vector sets, f i _3D is m...

Embodiment 3

[0160] Embodiment 3 experimental result

[0161] The point cloud model and RGB image used in the experiment of the present invention come from K.lai et al. (RGB-D dataset.http: / / rgbd-dataset.cs.washington.edu / dataset.html, 2011-03-05. For K.Lai, L.-F.~Bo, X.-F~Ren, D.~Fox, A Large-Scale Hierarchical Multi-View RGB-D Object Dataset, Proc.of IEEE Int.Conf.on Robotics and Autom., pp: 1817--1824, Shanghai, China, 2011.) established a large point cloud database, which contains 51 categories of point cloud models and RGB images of 300 objects, and each object point cloud and image contains 3 perspectives. Experimental method: randomly select one object in each category as a test sample, and the remaining objects as a training sample, select 100 training samples for each category, and 60 test samples, all randomly selected from the database. In order to evaluate the performance of the algorithm proposed in this paper, a number of experiments were carried out in this part, and the c...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an ordinary object recognizing method based on 2D and 3D SIFT feature fusion. The ordinary object recognizing method based on 2D and 3D SIFT feature fusion aims to increase the ordinary object recognizing accuracy. A 3D SIFT feather descriptor based on a point cloud model is provided based on Scale Invariant Feature Transform, SIFT (2D SIFT), and then the ordinary object recognizing method based on 2D and 3D SIFT feature fusion is provided. The ordinary object recognizing method based on 2D and 3D SIFT feature fusion comprises the following steps that 1, a two-dimension image and 2D and 3D feather descriptors of three-dimension point cloud of an object are extracted; 2, feather vectors of the object are obtained by means of a BoW (Bag of Words) model; 3, the two feature vectors are fused according to feature level fusion, so that description of the object is achieved; 4, classified recognition is achieved through a support vector machine (SVN) of a supervised classifier, and a final recognition result is given.

Description

technical field [0001] The invention relates to a general object recognition method based on fusion of 2D and 3D SIFT features, belonging to the technical field of recognition methods. Background technique [0002] General object recognition is a hot topic of research at home and abroad in recent years. It is different from specific object recognition (Specific Object Recognition), such as face recognition, which can be trained through a large number of training samples, and only deal with certain objects or types of objects; Object recognition is much more difficult, because it is necessary to use general features that are common among object classes, but not to define features for a specific category, and this feature needs to express the commonality and differences between classes as much as possible, and it must be able to handle multiple classes Classification and incremental learning, under this premise, it is impossible to use a large number of samples of a given cate...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/46G06K9/62
Inventor 李新德刘苗苗徐叶帆
Owner SOUTHEAST UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap