MOMENT BASED OBJECT RECOGNITION AND RECONSTRUCTION

In this book, the problem of object recognition and reconstruction is addressed using image moments. We discussed a Artificial Neural Network (ANN) based scheme for Object recognition. The invariance properties of geometric moments as well as lower order moments corresponding to partially occluded objects are used to train a feedforward ANN. The trained neural network is used to predict the actual moments from the moments of the object with different occlusions. The object is recognized based on the comparison of the actual moments and the predicted moments. The problem of object reconstruction using Zernike and Pseudo- zernike moments is also discussed.