Fast transform algorithm application method and apparatus for multi-scale auto-convolution

A technology of rapid transformation and application method, applied in the field of target recognition, which can solve the problems of high computational complexity, high time consumption and high computational complexity

Inactive Publication Date: 2013-09-11
RECONNAISSANCE INTELLIGENCE EQUIP INST OF ARMAMENT ACADEMY OF THE PEOPLES LIBERATION ARMY AIR FORCE
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although the calculation speed of the MSA method can be improved by fast Fourier transform, four Fourier transforms need to be completed when acquiring features for each pair of scales, which is computationally complex and time-consuming
Afterwards, based on multi-scale self-convolution transformation, methods such as normalized affine moment

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
  • Fast transform algorithm application method and apparatus for multi-scale auto-convolution
  • Fast transform algorithm application method and apparatus for multi-scale auto-convolution
  • Fast transform algorithm application method and apparatus for multi-scale auto-convolution

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0019] The present invention analyzes the computational complexity of the MSA transform in detail, and finds the key link restricting its computational speed: Fourier transform size and transform times. Based on the scale value range and characteristics, this is a breakthrough to improve the calculation speed of MSA. When the scale takes a uniform value in the smallest plane area, according to the self-derived minimum size requirement of the MSA transform, the minimum transform size that meets all scale requirements is calculated, effectively reducing the total number of Fourier transforms. Compared with the original method, the proposed new method has the same eigenvalue accuracy, and the calculation speed is about doubled. When the absolute value of the scale is greater than 1, scale transformation is carried out with the help of multi-scale self-convolution transformation properties, which greatly reduces the size of the Fourier transform. Computational speed increased by ...

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 a fast transform algorithm application method and an apparatus for multi-scale auto-convolution. The method includes the steps of reducing the number of the transformation of the multi-scale auto-convolution or transform sizes, establishing a multi-scale auto-convolution fast transform algorithm and using the multi-scale auto-convolution fast algorithm for target identification, matching, image registration or image retrieval. Based on the characteristic that values of scales within the smallest plane area are uniform, a minimum transform size requirement that MSA (measurement system analysis) value has nothing to do with the transform size is solved and proved, images of various scales are unified to the same size to be transformed, and thus, transformation times are greatly reduced, and transformation speed of MSA is improved. When maximum absolute values of the scales are more than 1, the scales are transformed within the range of (-1, 1) according to the transformation properties of the multi-scale auto-convolution, and thus, transform sizes are reduced, and the transformation speed of MSA can be effectively improved through the two ways.

Description

technical field [0001] The invention relates to the technical field of target recognition, in particular to an application method and device of a multi-scale self-convolution fast transformation algorithm. Background technique [0002] Object recognition is one of the key research topics of computer vision, which essentially uses some mathematical tools or models to extract invariants. In recent years, the extraction of invariants has been developed from a single target image to a target multi-scale image, and the ability to identify invariants has been significantly improved, for example, multi-scale Fourier descriptors, global affine invariants, binary wavelet affine invariants are better than the corresponding single-scale invariants. When the target size is much smaller than the imaging distance, the affine transformation model is the best choice to simulate the relationship between the images formed by these targets. Therefore, the use of multi-scale methods to extract...

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
IPC IPC(8): G06T1/00
Inventor 黄波庞怡杰陈东时公涛贺鹏陈涛张利宁
Owner RECONNAISSANCE INTELLIGENCE EQUIP INST OF ARMAMENT ACADEMY OF THE PEOPLES LIBERATION ARMY AIR FORCE
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