Traditional video monitoring mainly depends on playing back the original video,and takes no account of the night monitoring.Current night-vision monitor systems essentially apply infrared(IR) technology and image-enhanced technology to improve the quality of night-vision images.A video capture and display system is built on the platform of the High-performance DSP,and the lower video image mining is accomplished by embedding the Zadeh-X transform algorithm and the best mining parameters(sita and delta) are obtained automatically based on comentropy to achieve real-time mining.The results show that the images undistinguishable to human vision are reproduced and both the visual effects and the quality of the images in night real-time monitoring video are significantly improved.

Automatic parameter optimization for the lower layer image mining technique by virtue of contrast resolution limitation of human vision is proposed.In lower layer image mining based on Zadeh-X transformation,two optimum mining parameters are necessary:Sita and Delta.First,information entropy and average contrast are selected,which are related to the image quality of visual perception.Secondly,the function association of information entropy and average contrast to image mining parameter is studied.Then the relationship between the characteristis of trhe functions and subjective evaluation results is studied and no-reference image quality evaluation function is established based on information entropy and average contrast.Finally,by virtue of the criteria of the best quality image,parameters of Sita and Delta corresponding to the best quality image are quested to realize the automatic optimization of mining parameter.

This paper expatiates upon the basic algorithm of lower layer image mining for stream media image based on Zadeh-X and the design of lower layer image mining filter.The algorithm is realized under DirectShow platform.From the result,the filter is practicable,and can well realize lower layer image mining for loacal video file.

The existing night-vision system is mainly by virtue of image intensifier to improve image quality,but costs too much.Based on the Zadeh-X transformation,night vision video image mining was actualized by DirectShow develop platform besides.The new method can acquire the images that there are much betterment at level of images and the quality of image that is apperceived by human vision through two gray spectrum flattening analysis,achieving better results in the night vision video images mining.It will be an effective tool for post processing of the traditional night vision system.

The two-level inverted pendulum system based on fuzzy control is designed in the paper It is impossible to adopt conventional control theory in time, for the mathematical modeling of the system is nonlinear the fuzzy control is proposed to improve the former linear control algorithms. For the key to fuzzy control is the fuzzy control rules, it is important to design suitable and effective fuzzy rules. Two methods, one based on experiences and the other based on system simulation, are presented to set up an effective rule data-base. The practice proves that the control system is of well stability.