Abstract: Text-based and content-based retrieval are considered as two main approaches of the image-based retrieval. Challenges faced by both approaches direct researchers toward combined approaches and semi-automatic retrieval by the collaboration of the user in the cycle of retrieval specifically in the field of medical issues. Therefore, in this study, a system of image retrieval is introduced which provides user with two types of query based on keywords and images. After retrieval of the initial results, proposed system performs retrieval operation in a semantic manner and semi-automatically using feedbacks received from the user and high level semantic tags allocated to the images. By means of a hierarchical ST and a type of learning by user feedbacks, system is able to respond to diverse queries in the field of image retrieval. According to tests, proposed system is of acceptable precision.
Nasibeh Esmaeili Shahmirzadi and Ibrahim Chaoudi, 2018. Content-Based Image Retrieval Based on High Level Labels. Asian Journal of Information Technology, 17: 98-113.