IMAGE RECOGNITION USING MORPHOLOGICAL OPERATIONS
Ahmed Mahmoud Hammad;
Abstract
Image matching using morphological operations is a research point that combines the pattern recognition and image morphological processing which can successfully be used in pattern classification tasks. A technique for investigating and studying the problem of estimating the similarity between images is presented in this thesis. The proposed technique is applied to the similarity measure of both the binary and gray scale images. The shape representation for this new method utilizes morphological image opening and closing of the input image at different scales of a primary and rotated structuring elements. The areas of the opened and closed images are computed for each scale of the structuring element to construct the feature vector of each input image. This feature vector is called morphological descriptor.
The drawback of shape representation techniques based on processing single complex shape is that the extraction of complex shape properties may not be efficient. So, the proposed shape representation technique described is based on the decomposition of a complex shape to multiple simple signature shapes. The idea of this approach is to process decomposed, multiple shapes instead of processing the original shape.
The distance function based on the feature vector extracted using morphological operations called morphological signature transform distance is presented. Because the operations involved in morphological processing relate directly to its content, it is compatible with human visual system. So, the proposed technique is insensitive to some changes on images such as rotation, noise, scale, shifting columns to right, and filtering .
Most of the face recognition systems have difficulties in recognizing oriented (tilted) heads. The proposed matching technique presented in this thesis is applied to face recognition that concentrate on recognizing the faces of right head orientation, left head orientation, and facial expressions such as smiling faces. A collection of face images of IITKanpur university database is used. It is The proposed shape matching technique presented in this thesis is very sensitive to head orientation and smiling.
The drawback of shape representation techniques based on processing single complex shape is that the extraction of complex shape properties may not be efficient. So, the proposed shape representation technique described is based on the decomposition of a complex shape to multiple simple signature shapes. The idea of this approach is to process decomposed, multiple shapes instead of processing the original shape.
The distance function based on the feature vector extracted using morphological operations called morphological signature transform distance is presented. Because the operations involved in morphological processing relate directly to its content, it is compatible with human visual system. So, the proposed technique is insensitive to some changes on images such as rotation, noise, scale, shifting columns to right, and filtering .
Most of the face recognition systems have difficulties in recognizing oriented (tilted) heads. The proposed matching technique presented in this thesis is applied to face recognition that concentrate on recognizing the faces of right head orientation, left head orientation, and facial expressions such as smiling faces. A collection of face images of IITKanpur university database is used. It is The proposed shape matching technique presented in this thesis is very sensitive to head orientation and smiling.
Other data
| Title | IMAGE RECOGNITION USING MORPHOLOGICAL OPERATIONS | Other Titles | التعرف على الصور باستخدام المورفولوجي | Authors | Ahmed Mahmoud Hammad | Issue Date | 2005 |
Recommend this item
Similar Items from Core Recommender Database
Items in Ain Shams Scholar are protected by copyright, with all rights reserved, unless otherwise indicated.