Research Of Flexible Object Modeling Based On Active Shape Model

Flexible object can vary over time within a certain range of any deformation whichhas variability, unknown, universality and certain restrictive. We often need deal with theprocessing of flexible object image in computer vision and image processing. Flexibleobject has its own unique properties. It is more complex in the computer modeling andanalysis than the rigid object, so the corresponding model must have typical characteristicsand able to deal with the complex and changeful object.This paper researches active shape model which has the generality and specificity asone of the most effective way of analysis of the flexible object. The advantage of thismodel is that it can restrict the adjusting of parameters, so the chang in shape is controlledwithin a reasonable range. To solve the problem of the time consuming in the labeling ofthe training images, this paper extracts the edge by using multi-scale wavelet transformmodulus maxi-ma method, gets the threshold by maxing the between-class variance andfinally uses the connectivity judgment for each scale edge fusion. To solve the dilemma ofsearch profile selection, this paper puts forward multi-resolution search strategy. First,search for the best candidate for the approximate location of the point in the low resolutionof rough image and then make the improvement in the clear images which have higherresolution. In addition, this paper discusses the application of the algorithm in the field oftarget tracking. Aiming at the initial positioning of the model in the tracking process, thispaper has a detailed analysis of the principle of Kalman filtering and its target predictionand proposes a target tracking method based on the combination of Kalman filtering andASM algorithm. The introduction of the Kalman filter forecast in the target trackingprocess improved the efficiency of the contour tracking and made the initial positioning ofthe model more accurately.