Can a machine learn to see past the clutter? This paper introduces a novel method for recognizing occluded objects in images, leveraging the power of Markov models. The study demonstrates that Markov models, known for their high tolerance to noise, can be effectively adapted to incorporate spatial distribution of features in an image. Thus, resulting in improved recognition accuracy. More specifically, for each occluded object in the scene image, its translation, rotation and scale parameters can all be determined by our method even when it may have transformation parameters different from others or it may be duplicated in the scene image with transformation parameters different from each other. For each occluded object in the scene image, the study describes how its translation, rotation, and scale parameters can all be determined, even with varying transformation parameters or duplications. The recognition process proceeds step by step, identifying objects based on a confidence measure and terminating automatically without prior knowledge of the number of objects in the scene. These findings have potential applications in depth-search scenarios, such as printed circuit board inspection, underwater object searching, and underground mine exploration. Experimental results on puzzle and tool databases validate the effectiveness and practicality of the proposed approach, indicating its potential for real-world image recognition tasks.
This paper, published in the _International Journal of Pattern Recognition and Artificial Intelligence_, aligns perfectly with the journal's focus. By introducing a novel method for object recognition using Markov models, the article addresses a central theme within the fields of pattern recognition and AI, contributing valuable insights for researchers and practitioners in these areas.