发明名称 |
Recognition method and system for component images |
摘要 |
A recognition method includes: retrieving a component image from an image source; utilizing SVD (singular value decomposition) to decompose the component image; and utilizing the decomposed component image for image recognition. In an embodiment, the recognition method further includes: utilizing a light compensation method to calculate at least one light compensation coefficient; utilizing the light compensation coefficient to process the component image to obtain a light-compensated component image; and utilizing the light-compensated component image for image recognition. |
申请公布号 |
US8977045(B2) |
申请公布日期 |
2015.03.10 |
申请号 |
US201213549424 |
申请日期 |
2012.07.14 |
申请人 |
National Kaohsiung University of Applied Sciences |
发明人 |
Wang Jing-Wein |
分类号 |
G06K9/00 |
主分类号 |
G06K9/00 |
代理机构 |
Kamrath IP Lawfirm, P.A. |
代理人 |
Kamrath Alan D.;Kamrath IP Lawfirm, P.A. |
主权项 |
1. A recognition method comprising:
retrieving at least one image of an electronic component from an image source, with the at least one image having pixel values of color channels; searching at least one difference among the color channels of the at least one image; utilizing a singular value decomposition (SVD) method to process the at least one image to obtain a SVD matrix, wherein the SVD matrix is calculated using the at least one difference; applying a color space transformation (CST) method to execute calculation on each color channel, wherein the CST method comprises combining color coefficients by computing a product of the SVD matrix of each of the color channels to obtain a CST image; and projecting the CST image to a high-dimension space for generating correlation and differences of information of the at least one image; and recognizing at least one electronic component from the projected CST image. |
地址 |
Kaohsiung TW |