Hierarchical Implicit Shape Modeling

In this paper, a new hierarchical approach for part-based object recognition is proposed. Object detection methods based on Implicit Shape Model (ISM) efficiently handle deformable objects, occlusions and clutters. The structure of each object in ISM is defined by a spring like graph, hence parts independently vote to object properties. We introduce hierarchical ISM in which structure of each object is defined by a hierarchical star graph. The proposed approach has two layers. In the first layer, extracted low-level features vote to the center of mid-level parts of the object. Mid-level parts are not labeled in training samples and are treated as latent variables. In the second layer, the mid-level parts vote to the object’s centroid. The obtained mid-level parts have local characteristic and discriminative power in object class recognition. Therefore, the proposed approach does not require a verification stage. We applied the proposed approach to some known datasets and compared the performance of our algorithm to comparable methods available in the literature. The results of our experiments show that our method has a superior or comparable performance to the most successful and recent existing algorithms. Keywordsstatistical part-based object recognition, discriminative parts, Hierarchical Implicit