HSV Color Histogram Based Image Retrieval with Background Elimination

In this study, a new content based image retrieval (CBIR) method, which uses HSV histogram data is proposed. The model uses the HSV histogram to find the background from the image by analyzing the peaks in the histogram data and performing a moving window algorithm to identify the region within the histogram that belongs to the background colors. After identifying the background information, the sections of the image that are part of the background are removed from the original image and the remaining foreground or content information is stored for comparison with other images. In order to verify the methodology, a graphical user interface is developed and 1000 different images from 10 different groups from the coral database are put into the image database for comparison. The analysis and preliminary tests show that comparing only the foreground information of the images pro- vided better results than comparing images themselves, especially when searching for particular objects within the images. This algorithm can also be used as a background elimination technique to reduce the storage requirements of images and the comparison time between images can be reduced significantly.