A corner feature adaptive neural network model for partial object recognition

Real Time object recognition is the major classification method used in many authentication and recognition applications. But it becomes difficult to identify the object when the input object image is not complete. In this paper, partial object recognition is presented based on the corner point effective mapping. The work is here divided in two main stages. In first stage, the input object and dataset images are transformed to featured form using corner point analysis. Later on neural network classifier is applied to perform the classification. The work is implemented in matlab environment on different sample sets. For each sample set, the over 90% recognition rate is achieved.