Content-based image retrieval system for marine invertebrates

Abstract

There are many marine life around the world where it is essential to have proper documentation for future records. Many information retrieval systems for marine science today require text input from user and can only be accessed online. Therefore, people who do not know the name of the marine species or do not have Internet access cannot search using the systems. Responding to this important need, this work aims to develop a Content-based Image Retrieval (CBIR) system for marine invertebrates based on colour and shape features. With the CBIR system for marine invertebrates, users can use the system to look for marine invertebrates' species instead of using traditional methods of searching such as using books and encyclopedias. Users can easily upload the image of marine invertebrate that they want to search into the system and the system will retrieve all the other similar images of marine invertebrates along with their description. All the system interface's buttons, icons and text were designed in a way where any user can easily understand and further learn to operate the system themselves. Based on the retrieval effectiveness experiment and questionnaire-based survey, the proposed CBIR system for marine invertebrates is shown to be effective, help users search similar images of marine invertebrates, provide concise information on marine invertebrate's species for learning purposes, and is reliable and user-friendly.

Content-Based Feature Fusion Representation for Marine Invertebrates

“…It is very important to determine the number of order so that images will be represented well according to the optimum number of repitition. Second experiment is to determine the retrieval effectiveness of the proposed feature fusion method on the marine life image dataset compared to the work done in [10] and few other benchmark methods.…”

“…Colour and shape features have been considered for describing marine invertebrate images in [10]. Colour Moments is used to represent colour information while Discrete Wavelet transform and Canny edge detector are applied to extract shape details.…”

“…It provides great result. We are extending the work done in [10] to include texture features so that more effective method can be developed for marine life content-based image representation and retrieval. Malaysian Journal of Computer Science, Vol.…”

“…All images are used as query. This dataset is similar to the one used for our earlier works mentioned in [10]. The ground truth of this dataset is based on the 10 similar images within a class.…”

“…We tested on the order n = 0 up to the order of n = 10. We conducted the evaluation on 50 marine invertebrate images which have been used in our previous experiment on marine life [10]. All 50 images are used as query.…”