Method of segmentation of a digital image and a method of controlling an implement
First Claim
1. A method of segmenting a digital image divided into small elements (pixels) and acquired with a camera having several colour channels, to discern an object from a background, in which method,a priori, a daylight locus and a colour locus of the camera are modelled in a space defined directly by the colour channels of the camera,a daylight locus being the response of the camera to a standard daylight of varying correlated colour temperature (CCT), andthe colour locus being modeled after body reflectance characteristics of the object under conditions of said standard daylight, wherein,a priori, a normal vector (ns) to the modelled daylight locus is determined and for each small element, the scalar product between a colour vector (Cf) of the element in the form of the response from the colour channels of the camera and the normal vector of the daylight locus is calculated, following which the element is characterized as representing object or background on basis of the calculation result, the elements characterized as objects being assigned one value and the elements characterized as background being assigned a second value.
1 Assignment
0 Petitions
Accused Products
Abstract
In connection with the segmenting of a digital image divided into small elements (pixels) and acquired with a camera having several color channels, to discern an object from a background, the daylight locus and the color locus of the camera are, a priori, modeled after to the body reflectance characteristic of the subject. In this way a normal vector (nS) to the daylight locus is determined a priori and for each pixel, the scalar product between a color vector (cf) of the element in the form of the response from the color channels of the camera and the normal vector of the daylight locus is calculated, following which the element is characterized as representing object or background on basis of the calculation result, the elements characterized as objects (1) being assigned one value and the elements characterized as background being assigned a second value. On basis of the segmented image a control signal for an implement (4) is derived.
13 Citations
3 Claims
-
1. A method of segmenting a digital image divided into small elements (pixels) and acquired with a camera having several colour channels, to discern an object from a background, in which method,
a priori, a daylight locus and a colour locus of the camera are modelled in a space defined directly by the colour channels of the camera, a daylight locus being the response of the camera to a standard daylight of varying correlated colour temperature (CCT), and the colour locus being modeled after body reflectance characteristics of the object under conditions of said standard daylight, wherein, a priori, a normal vector (ns) to the modelled daylight locus is determined and for each small element, the scalar product between a colour vector (Cf) of the element in the form of the response from the colour channels of the camera and the normal vector of the daylight locus is calculated, following which the element is characterized as representing object or background on basis of the calculation result, the elements characterized as objects being assigned one value and the elements characterized as background being assigned a second value.
Specification