Method of detecting change using image

• US 4,912,770 A
• Filed: 11/04/1987
• Issued: 03/27/1990
• Est. Priority Date: 11/07/1986
• Status: Expired due to Term

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Seto, Youichi, Komura, Fuminobu
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): COUSO, YON JUNG
• Application Number:
  US07/117,286
• Time in Patent Office: 874 Days
• Field of Search: 382/44, 382/17, 382/1, 382/6, 358/105, 358/109
• US Class Current: 382/107
• CPC Class Codes: G06T 7/254 involving subtraction of images

Abstract

A method of detecting a change using an image includes an image inputting section, an image registration processing section, a binarization processing section and an image display section. In this method, images taken at different times are input, and any change in corresponding pixels in the input images are detected. The change detecting method includes a step of calculating differences in vectors of a plurality of images which are based on different image characteristics, and a step of extracting values of shape features of an object, so that the direction in the change that takes place in the image characteristics of the object is calculated for each image characteristic being observed, shape feature values of the object are calculated, and the calculated direction and shape feature values are compared with corresponding training data, so as to detect the change in the object.

Claims

1. A method of detecting a change using input images taken at different times by detecting any change in corresponding pixels representative of said input images, said change detecting image method including an image inputting section, an image registration processing section, a binarization processing section, and an image display section, wherein the method comprises:

a step of calculating differences in vectors of a plurality of images which are based on different image characteristics, the vectors being representative of a change in image characteristics of various objects in the image data and classifying the image characteristics into statistical categories such that candidate areas are extracted by calculating a change in direction in the vectors of an object and by comparing said direction with predetermined training data, and conducting a statistical category classification process on said candidate areas alone so as to detect the change in said object.

2. A change detection method using an image according to claim 1, wherein said plurality of images which are based on the different image characteristics are multispectral images, and said image characteristics are wavelength characteristics.

3. An image analyzing system for detecting a change in an object from multitemporal images, comprising:

a first means for inputting first image data obtained by taking a picture of said object at a certain time;

a second means for inputting second image data obtained by taking a picture of said object at a time different from said certain time;

a third means for performing registration on said first and second image data;

a fourth means for calculating the magnitude of difference and direction of change in vectors representative of a change between said first and second image data on which registration has been performed;

a fifth means for storing threshold values obtained from known detail data on said object;

a sixth means for detecting the change in said object from the relationship between said magnitude of said vector difference and said direction of change with said threshold value; and

,

a seventh means for adjusting levels of brightness so as to correct the atmospheric conditions of said first and second image data.