Penerapan Deskriptor Warna Dominan untuk Temu Kembali Citra Busana pada Peranti Bergerak

Yustina Dhyanti, Khairul Munadi, Fitri Arnia

Abstract


Nowadays, clothes with various designs and color combinations are available for purchasing through an online shop, which is mostly equipped with keyword-based item retrieval. Here, the object in the online database is retrieved based on the keyword inputted by the potential buyers. The keyword-based search may bring potential customers on difficulties to describe the clothes they want to buy. This paper presents a new searching approach, using an image instead of text, as the query into an online shop. This method is known as content-based image retrieval (CBIR).  Particularly, we focused on using color as the feature in our Muslimah clothes image retrieval. The dominant color descriptor (DCD) extracts the wardrobe's color. Then, image matching is accomplished by calculating the Euclidean distance between the query and image in the database, and the last step is to evaluate the performance of the DWD by calculating precision and recall. To determine the performance of the DCD in extracting color features, the DCD is compared with another color descriptor, that is dominant color correlogram descriptor (DCCD). The values of precision and recall of DCD ranged from 0.7 to 0.9 while the precision and recall of DCCD ranged from 0.7 to 0.8. These results showed that the DCD produce a superior performance compared to DCCD in retrieving a set of clothing image, either plain or patterned colored clothes.

Keywords


muslimah fashion; CBIR; feature extraction; DCD; recall and precision

Full Text:

PDF


DOI: https://doi.org/10.17529/jre.v12i3.5701

Refbacks

  • There are currently no refbacks.


View My Stats