Object Recognition and Clustering based on Latent Semantic Analysis (LSA)

Download Now
Provided by: Creative Commons
Topic: Data Management
Format: PDF
Object recognition and clustering are prime techniques in computer vision, pattern recognition, artificial intelligence and robotics. Conventionally these techniques are implemented in visual-feature based methods. However, these methods have drawbacks they do not efficiently deal with the differences in shapes and colors of objects. Another method which uses semantic similarity to solve this kind of problem, i.e. cosine similarity method, but this method also has problems. The problems are synonymies and polysemies. In this paper, the authors propose a method in which objects with different shapes and different colors which function similarly can be recognized and clustered.
Download Now

Find By Topic