RECOGNITION OF PARTIALLY OCCLUDED INDUSTRIAL OBJECTS USING EXPECTATION MAXIMIZATION

This paper is concerned with the partially industrial occluded object recognition using Expectation Maximization Mixture of Gaussian (EM-MoG) algorithm. The proposed method takes an input image containing occluded objects and extracts pixels features then uses EM-MoG algorithm to label the occluded input image. After the labeling process, the proposed method uses a two stages matching process to identify the objects. First, it compares the components of the labeled image with a database of template object images to find the most probably matched component. Second, it applies logic subtraction and combination step to reconstruct an object which is matched again with templates to ensure the final result. The proposed method has been implemented and tested in MATLAB environment. The experimental results demonstrate the accuracy and robustness of the proposed method in comparison with other methods