Automatic image annotation using an ant colony optimization algorithm (ACO)
Kavita Akhilesh, Raghavendra R. Sedamkar · 2016
Automatic Image Annotation (AIA) for a large collection of images is one of the most challenging topics for researchers in the past years wherein, the important task for researchers is a region labelling, since the whole feature does not give useful information for each conception. Feature selection is considered to be the important pre processing method because it may affect the performance of image classification. Each input image is first segmented using mean shift algorithm which partitioned the original image into many segments followed by feature extraction step, it is considered to be an important preprocessing method which contains the optimization of feature descriptor weights. ACO is an iterative, probabilistic, meta-heuristic method and can be employed as a search strategy to identify an optimal feature subset. Hence, it improves the performance of image annotation by reducing the feature dimension. Our proposed hybrid approach (Image segmentation and ACO) not only deals with the region labelling but also finds tags for a query image.