The Video Defense Model On The Body Image Based On The Early Warning

Organizing a defense divided into two models: one is parameter defense, another is non-parameter defense.The parameter in the parameter defense figures four kinds: form feature; color feature; textural feature and spatial feature. The object divided into two parts that is early warning to object and goal to object. The core conception of parameter defense is as follows 1) the early warning to the feature of object; 2) the definite to feature parameter; 3) related algorithm of feature parameter; 4) similarity analysis.Traditionally, the non-parameter defense is based on method of lattice. The core conception of defense is by the way of auto fixing alarming region, dividing the monitor pictures into different regions, noticing the trivial changes between the two adjoining region to see whether there is an alarming. And the way to accumulate is easy. In some extend, it can satisfy the need of monitoring and alarm in time when different changes happened in the region.But the disadvantage is also obvious, presenting as following points: 1)bad immunity; 2)low alarm accuracy; 3) regular defense region; 4)lacking the necessity dealing with the alarm object information, making the after-function lose the fundamental for lower intelligence.The article makes a research on the definite and extraction of the parameter in the parameter defense model. Firstly, we filter the video pictures which were collected and get rid of the inclusion; secondly, examine the specific color of the object, making sure the ROI region according to the color feature, then, making a specific parameter description of the defense object. Form feature parameter description adopts the Fourier changing model which uses the Fourier to descript the changing of Fourier on the board, make the feature description as the shape. Color feature parameter description uses RGB-HSZ model, eliminating the workforce of picture analysis.Textural feature parameter description uses grey level-element co-exist matrix, synthesizing vital statistics of picture with structural feature, making a better description about textural structural to the object. Spatial relation feature description employs cognitive map. The cognitive map was formed with symbol, void build with back-up, network node and the link band between the links. Every network node has its state value, portraying the nature of the environment goal. Lastly, we are making the extraction of the feature parameter to the object. The extraction of form feature parameter adopts the small wave and relative square and its advantage is that it can describe the square model by using multi-scale form description. Color feature parameter extraction uses grey level co-exist matrix, expressing the feature of the pictures about some related relation by using pictures grey square. Spatial relation feature parameter extraction drawing one or more controlling points from the two pictures, seeking the best changing model number of the controlling point in the space to some extend and match up with the controlling points.Through certain algorithm carries on the description and the extraction after the object body characteristic, establishes the similarity model, The experiment indicated that the similarity match effect is good, the algorithm operation efficiency was high, raised the efficiency which the parameter deployed troops for defense, entire deployed troops for defense process robust.