| 摘要 |
Disclosed are a method and a device for sensing smoke by using a sub-band energy ratio in a wavelet transformation domain. The method of the present invention comprises the steps of: transforming an RGB input image into a brightness image I (t); generating a binary image G1 having a background image and an entity image by comparing a first critical value with a difference between a current brightness image I (t) and a previous brightness image I (t-1); dividing an MxN image corresponding to the entity image of the binary image G1 into a particular number of blocks, performing first stage wavelet transformation on the divided blocks, and calculating DC sub-band energy E_L and high frequency sub-bad energy E_h; and sensing smoke in the entity image based on a ratio between the DC sub-band energy E_L and the high frequency sub-band energy E_H and a ratio between the DC sub-band energy E_L and the high frequency sub-band energy E_H in a continuous image. According to the present invention, by using an energy value for each sub-band of a wavelet coefficient obtained through wavelet transformation and relationship therebetween, a moving object and smoke can be reliably separated. |
