Machine vision-based network monitoring system for solar-blind ultraviolet signal

Abstract

The ultraviolet (UV) wavelength of 240 nm to 280 nm is called the solar-blind area. Therefore, if the UV signal in the solar-blind area can be detected, then it can only be the radiation from the measured object on the earth. To meet the requirements of UV signal detection, such as corona discharge and forest fire, this work designs a kind of UV dual spectrum imaging monitoring system. This system uses visible and UV dual light path structure for imaging, selects UV narrow-band filter to obtain UV light, and utilizes spectral conversion and image enhancement technology to image it. Web server and digital signal processing realize the function of remote network monitoring, and users can monitor and detect objects remotely through the client. An image fusion method based on nonsubsampling contourlet transform (NSCT) and visual saliency is proposed. Experimental results show that the fusion effect based on subjective visual effect and objective evaluation criteria is good. This method can also obtain high standard deviation, information entropy, edge preservation, and mutual information and preserve the details of the fusion image effectively. The results of system running and debugging show that the system design and image processing scheme proposed in this work are effective.