WO1997018523B1

WO1997018523B1 - Computer stereo vision system and method

Info

Publication number: WO1997018523B1
Application number: PCT/IL1996/000145
Authority: WO
Filing date: 1996-11-12
Publication date: 1997-08-21

Abstract

The system and the method are intended to enable a robot (20) to see the environment in which it operates, by means of a pair of identical cameras (3). The pictures from the cameras (3) are processed by a coordinator/translator (59). The system includes a memory register (51) for movement identification at time intervals.A basic shapes registry (52) identifies shapes.

Claims

-32-

AMENDED CLAIMS

[received by the International Bureau on 2 July 1 97 (02 07 97), original claims 1 and 9 amended, remaining claims unchanged (3 pages)]

1. A stereo computer vision system including: a. A pair of identical, aligned, coordinated cameras at fixed distance M, with coordinated photographing/filming angle and enlargement/ reduction, creating for the photographing cameras optical parallel fields of sight, coordinated, aligned, with identical field of sight line in both cameras from (0:0) and up to (Y:X), at a fixed distance M at any photographing distance, and the pictures of said cameras are received by means of input-memory devices in the computer, translated to computer language; b. Based on step A, the pixels of the received pictures are matched by the available means, at photographing rate and/or at any other rate to each of the previous picture(s) existing in the system's spatial memory registry according to the coordinates and space-counters, and when no matching is found there, the system performs updating and proceeds with matching of two pictures that have been input; c. Based on steps A and B, with regard to unidentified areas or areas in which any movement takes place and which are seen simultaneously in both cameras, the system looks for various features of the picture, such as color, and based on the difference in the count of pixels in each picture taken by a camera from the beginning of a line up to that matching point, calculation of the distance and the size of a point in space, represented by a pixel, are performed, and based on this, also the size of the area, the distances, etc.; d. Based on steps B and C, in matching points of unidentified areas with stored pictures, there will be points around the area that will not match due to a change in the distance between the area and the surrounding environment and the distance between the two cameras, these depth dots creating a contour-line and/or dividing line in addition to the approximately similar distance to the area dots, and with regard to areas at similar distance, changes in color or shade between areas and/or area motion/movement, will allow, by means of -33-

infoπnation; said means for receiving, transmitting and drawing of data; said means including power supply, data protection, protection of said means, the system and alike.

8. A computer vision system as in any of claims 1 to 7, wherein the received and stored information that has been collected before, during and after identification, including calculations, known data, features, definitions that have been collected such as with regard to areas, shapes, is immediately or after a short while transmitted forward in form of information, in a regular manner and/or by stereo in form of multimedia and/or in 3D form, and which the system preserves and/or provides and/or which are accessible upon request and/or automatically to the user, such as a robot, a device, a blind person, by means of adequate interfacing means.

9. A stereo computer vision method including: a. A pair of identical, aligned, coordinated cameras at fixed distance M, with coordinated photographing/filming angle and enlargement/ reduction, creating for the photographing cameras optical parallel fields of sight, coordinated, aligned, with identical field of sight line in both cameras from (0:0) and up to (Y:X), at a fixed distance M at any photographing distance, and the pictures of said cameras are received by means of input-memory devices in the computer, translated to computer language; b. Based on step A, the pixels of the received pictures are matched by the available means, at photographing rate and/or at any other rate to each of the previous picture(s) existing in the system's spatial memory registry according to the coordinates and space-counters, and when no matching is found there, the system performs updating and proceeds with mater ing of two pictures that have been input; c. Based on steps A and B, with regard to unidentified areas or areas in which any movement takes place and which are seen simultaneously in both cameras, the system looks for various features of the picture, such as color, and based on the difference in the count of pixels in each picture taken by a camera from the beginning of a line up to that matching point, calculation of the distance and the size of a point in space, represented by a pixel, are performed, and based on this, also the size of the area, the distances, etc.; d. Based on steps B and C, in matching points of unidentified areas with stored pictures, there will be points around the area that will not match due to a change in the distance between the area and the surrounding environment and the distance between the two cameras, these depth dots creating a contour-line and/or dividing line in addition to the approximately similar distance to the area dots, and with regard to areas at similar distance, changes in color or shade between areas and/or area motion/movement, will allow, by means of the available means, to detect and define a frame or countour for each and every area; e. For purpose of collection of additional data enabling to perform with the means available calculations, matching, setting of definitions and drawing of conclusions, the picture received from one of the cameras undergoes color filtering and is entered into the movement- identification register at time-intervals, after being duplicated and is then examined at fixed time intervals and based on steps B to D, the frame or countour of the area that has been detected is matched in size with the basic-shapes register, said data from these registers enable to calculate, for example, the size of an area/shape and speed of movement/motion in time and additional data, part of said data that have been input and part of the data that have been calculated/ detected previously allow matching as against a register of table(s) such as "true" for setting definitions and drawing conclusions with regard to other key-elements, such as: inanimate object, fluttering, animate, for example, if the size of the area goes from size to size and there is movement and/or motion in the envelope and/or inside it, etc.; f. Said key-elements drawn from data, features, definitions and conclusions, ordered in a particular order, allow, through the means provided, to detect, match and identify areas as against a register of stored shapes; g. The unidentified areas will be stored under temporary names and the identified shapes will be stored under their own name and with reference to the coordinates data in each scanning of the first dot and