A wide range of vision tasks, including semantic segmentation and object recognition, can be significantly improved by boundary and edge information. The semantic edge detection problem is more difficult than classical edge detection, which is a difficult task in and of itself. Many computer vision tasks, including 3d reconstruction , 3d form recovery , medical image processing , and semantic segmentation [4,5], have demonstrated to benefit from the classical edge detection job. Two pre-processing methods for image/object retrieval are suggested in this research. The object is defined as a pseudo-time series in one dimension using a traditional pre-processing method. The first method that is offered modifies the SAX representation by using an Extended SAX (ESAX) approach to quickly and accurately identify key patterns, which is necessary to identify the related items that have the best chance of being true. The general resemblance between two families of symbolic words is then used to characterise the relationship between two images or objects. To determine the most likely matching between strings of symbolic words, a distance measure is employed. We empirically compare the Extended SAX strategy to the original SAX approach and show how it performs better at getting the most likely related objects at larger cardinality. The second method uses a specified collection (subset) of boundary points to identify each object or form. Each point is the centre of a small area that surrounds it and is represented by an image patch that reflects the patch’s low-level properties. A family of image/object features corresponding to the image/patches object’s is connected to it. The degree of general similarity between two families of image/shape patches is then used to define the similarity between two images or objects. To select the most likely matches, GA is used. The experimental findings have demonstrated that our approach is successful at retrieving related photos.
Hala Ahmed Abdul-Moneim,
Department of Mathematics, Darb University College, Jazan University, Jazan, Saudi Arabia and Department of Mathematics, Faculty of Science, Minia University, Minia, Egypt.
Please see the link here: https://stm.bookpi.org/COSTR-V5/article/view/8336
Keywords: 1-D representation of objects, symbolic aggregate approximation (SAX), shape number, time series, extended sax representation, GA