Method for image processing

• US 6,546,153 B1
• Filed: 03/20/2000
• Issued: 04/08/2003
• Est. Priority Date: 05/06/1997
• Status: Expired due to Fees

• Current Assignee: Finn Hoydal
• Original Assignee: Dimensions AS
• Inventors: Hoydal, Finn
• Primary Examiner(s): COUSO, YON JUNG
• Application Number:
  US09/423,191
• Time in Patent Office: 1,114 Days
• Field of Search: 382/285, 382/154, 382/276, 382/286, 702/157, 702/150, 348/47, 348/578, 348/580
• US Class Current: 382/285
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 7/90 Determination of colour characteristics

Abstract

A method for image processing, especially for converting a two-dimensional image of a three-dimensional real subject into a three-dimensional representation of the same three-dimensional real subject, wherein the subject is composed of elements, each of which is represented by a pixel in the two-dimensional image. The image'"'"'s colors are subjected to an analysis with regard to color quality, and the individual color points'"'"' orientation in space is localised by utilizing a color temperature scale, a color saturation scale and a contrast scale, with the result that each color obtains its perspective place relative to the image'"'"'s other colors.

Claims

1. A method for image processing, especially for converting a two-dimensional image of a three-dimensional real subject into a three-dimensional representation of the same three-dimensional real subject, wherein the subject is composed of elements in the two-dimensional image, and wherein the two-dimensional image is provided by means of a camera, characterized by:

defining a reference plane corresponding to the camera'"'"'s focal plane lying closest to the camera'"'"'s focal plane, the reference plane comprising elements corresponding to elements in the two-dimensional image,

defining colour parameters hue, saturation, brightness for each element in the reference plane,

creating a reference scale by determining values of the colour parameters by means of an image sequence of individual images, each of which depicts the subject in different predetermined focal planes, the changes in colour parameters between the respective focal planes being used for calibration of a geometric measuring scale which refers respective measured colour parameters to metric or perspective distance values,

measuring and registering colour parameters hue, saturation and brightness for each element in the two-dimensional image,

comparing colour parameters for each element in the two-dimensional picture with colour parameters for the corresponding element in the reference plane,

based on said comparison, assigning to each element in the two-dimensional image a distance value dh;

ds;

db, where dh results from comparing hue values, ds results from comparing saturation values and db results from comparing brightness values,

calculating the distance d(z) between the elements in the two-dimensional image and the reference plane, d(z) being the distance measured along a z-axis perpendicular to the reference plane and with origin therein, as a weighed average of the distance values dh,ds,db;

d(z)=⅓

*(kh*dh+ks*ds+kb*db), where kh, ks and kb constitute weight factors based on empirical data,

performing the transformations d(x)→

X, d(y)→

Y and d(z)→

Z, where X,Y,Z are the element'"'"'s co-ordinates in a three-dimensional co-ordinate system with origin in the reference plane, the transformations d(x)→

X, d(y)→

Y being obtained by means of a known procedure for converting image elements in the plane into elements in space by means of calculated ratios (scale) between the camera'"'"'s image plane and the focal planes in the space, the closest plane constituting the reference plane, and the transformation d(z)→

Z being obtained by reverse application of the procedure for frontal perspective construction for reading metric distance from a standardised perspective drawing, and

determining on the basis of the established distance value d(z) and the element'"'"'s x,y co-ordinates in the two-dimensional image plane real spatial co-ordinates X,Y,Z for the element.

2. A method according to claim 1, characterized in that the empirical data are organised in a geometrical colour model suitable for use in connection with frontal perspective construction.

3. A method according to claim 2, characterized in that the geometrical colour model'"'"'s measuring scale comprises geometrically coincident non-linear scales based on reference and empirical data for hue, colour saturation and brightness respectively, where the colour temperature increases towards its maximum value at the common end point, or point of convergence, where the colour saturation and the contrast approach zero.

4. A method according to claim 3, characterized in that the non-linear scales'"'"' common point of convergence is determined by means of the colour parameters for one or more remotely located elements in the image which was focused on the focal plane for furthest distance, the measured values for this element or average measured values for these elements determining the point of convergence.

5. A method according to claim 4, characterized in that the geometrical measuring scale is inserted in a frontal perspective diagram which corresponds to the format of the two-dimensional image and with the non-linear scales calibrated for the subject of the image concerned for determination of the distance between the reference plane and an element in the image.

6. A method according to claim 5, characterized in that a perspective-based distance value is employed.

7. A method according to claim 6, characterized in that a metric distance value is employed.