Background subtraction method based on color constant parameters

Abstract

The invention discloses a background subtraction method based on color constant parameters. By defining the color constant parameters directing at RGB (red green blue) images as main basis for distinguishing background pixels from foreground moving targets, the method solves the problem that foreground targets have holes due to color similarity in background subtraction technology. Besides, the method is efficient in running, small in occupied computation memory, short in consumed time, suitable for embedded application and capable of realizing real-time video monitoring.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a background subtraction method based on color invariant parameters. technical background [0002] Background subtraction is a basic problem in the field of computer vision, and it is the first problem to be solved in the image processing part of all computer vision applications. Video surveillance is an important application of computer vision, which is widely used in security, industrial control, automatic management and other fields. The quality of background subtraction technology directly determines the effectiveness of video surveillance. [0003] In the current background subtraction technology, the misjudgment of the foreground object as the background environment often occurs. An important reason for this error is that the background subtraction method used cannot be effective when the foreground object and the background environment have similar colors. Theref...

AI Technical Summary

Problems solved by technology

[0003] In the current background subtraction technology, the misjudgment of the foreground object as the background environment often occurs. An important reason for this error is that the background subtraction method used cannot be effective when the foreground object and the background environment have similar colors. Distinguishing background and foreground objects, thus causing misjudgment of foreground objects as background or background as foreground objects

The current mainstream background subtraction methods generally use more complex algorithms. Therefore, the operation of background subtraction is often time-consuming and consumes a large amount of computer memory. This limitation increases the difficulty of embedded implementation of video surveillance and increases the The cost of video surveillance

Images