公开(公告)号：WO2006000879A1

公开(公告)日：2006-01-05

申请号：PCT/IB2005/001754

申请日：2005-06-21

Applicant: LAMBDA SOLUTIONS, INC.

Inventor： LIU, Canwen

IPC: H04N5/21

CPC classification number: G01N30/8641 ,  G01N21/274 ,  G01N2201/12784 ,  G06T5/002

Abstract: Automatic background signal removal for input data, such as for spectrometry data, is provided. Input data includes input pixel points, such as those read by a CCD spectrometer or chromatography device, and intensity values corresponding to the data points. A distribution of changes in the intensity values between the data points is determined, and a noise level is judged by setting a threshold for the distribution. A noise region is identified as a predetermined number of consecutive input points for which the changes in the intensity values are within the noise level. Adjacent noise regions may be connected and the background signal is thus determined and subtracted. A spike noise region may also be identified and filtered, such that a peak obtained from fewer than a second predetermined number of the pixel points is determined as a spike, not a true peak. Non-spike large amplitude noises are optionally filtered.

Abstract translation: 提供输入数据的自动背景信号去除，如光谱数据。 输入数据包括诸如由CCD光谱仪或色谱装置读取的像素点和对应于数据点的强度值的输入像素点。 确定数据点之间强度值变化的分布，并通过设定分配阈值来判断噪声水平。 噪声区域被识别为强度值的变化在噪声水平内的预定数量的连续输入点。 可以连接相邻的噪声区域，从而确定并减去背景信号。 也可以识别和滤波尖峰噪声区域，使得从少于第二预定数量的像素点获得的峰值被确定为尖峰，而不是真峰值。 可选地过滤非尖峰大振幅噪声。

公开(公告)号：EP1782621A1

公开(公告)日：2007-05-09

申请号：EP05757723.1

申请日：2005-06-21

Applicant: Lambda Solutions, Inc.

Inventor： LIU, Canwen

IPC: H04N5/21 ,  G06T5/00 ,  G01N21/27 ,  G01N30/86

CPC classification number: G01N30/8641 ,  G01N21/274 ,  G01N2201/12784 ,  G06T5/002

Abstract: Automatic background signal removal for input data, such as for spectrometry data, is provided. Input data includes input pixel points, such as those read by a CCD spectrometer or chromatography device, and intensity values corresponding to the data points. A distribution of changes in the intensity values between the data points is determined, and a noise level is judged by setting a threshold for the distribution. A noise region is identified as a predetermined number of consecutive input points for which the changes in the intensity values are within the noise level. Adjacent noise regions may be connected and the background signal is thus determined and subtracted. A spike noise region may also be identified and filtered, such that a peak obtained from fewer than a second predetermined number of the pixel points is determined as a spike, not a true peak. Non-spike large amplitude noises are optionally filtered.