Unlock instant, AI-driven research and patent intelligence for your innovation.

A Local Feature Description Method

A technology of local features and local blocks, applied in computer parts, instruments, computing, etc., to achieve the effect of being conducive to transmission and storage, high application value, and excellent de-correlation

Active Publication Date: 2019-05-14
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Most of these existing methods increase the matching speed at the expense of matching quality

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
  • A Local Feature Description Method
  • A Local Feature Description Method
  • A Local Feature Description Method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0022] Such as Figure 1-4 As shown, the present invention includes several steps of preprocessing, polar grid sampling, mapping, DCT calculation, zigzag scanning and descriptor formation. Details are as follows.

[0023] Step 1. Preprocessing: feature detection and normalization.

[0024] Step 1: Use a feature detector to detect the local feature region of the image and map it to a local block of size d×d (d=41). Local feature region detection uses Hessian affine detector.

[0025] Step 2: Use Gaussian filtering to remove interpolation noise caused by affine normalization. Gaussian filtering uses a Gaussian kernel with a size of 5×5 and a standard deviation of 1.

[0026] Step 3: Use the mean μ and standard deviation σ of the local block gray value obtained in step 2 to normalize each pixel x of the block again i , remove the influence of illumination changes to get the normalized local block gray value y i , get as figure 1 The preprocessed local blocks are shown. the...

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 present invention provides a local feature description method, which firstly normalizes the obtained local interest area, then uses the polar coordinate sampling grid to divide the local block, performs quantitative sampling on the local block, maps it into a 2-dimensional matrix and extracts the 2-dimensional DCT frequency domain features, and then scan the DCT coefficient matrix in zigzag order, rearrange and screen DCT features to form the final local descriptor. The invention uses the polar coordinate sampling structure, not only retains the original space information of the local block, but also can tolerate certain deformation and enhance the robustness of the descriptor. The calculation of 2-dimensional DCT features is simple, efficient, and compact, and the filtered DCT features further remove the influence of illumination.

Description

technical field [0001] The invention designs a digital image processing technology, and particularly relates to a local feature processing technology of an image. Background technique [0002] Compared with the global features, the local features of the image can obtain reliable matching when the object is deformed or occluded, so it has attracted much attention in the fields of image processing and computer vision. Local features have been widely used in wide-baseline matching, object recognition, texture classification, image retrieval and other fields. [0003] In general, local features involve three steps: local feature detection, local feature description, and local feature matching. First use the feature detector to detect the interest point (or interest area) of the image, then calculate the image descriptor on the surrounding support area of ​​the located interest point, and finally use the appropriate matching function to measure the distance between the descripto...

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 Patents(China)
IPC IPC(8): G06K9/46
Inventor 李宏亮宋铁成
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA