Montage : An image database for the fashion, textile, and clothing industry in Hong Kong

Abstract

Abstract. The fashion, textile, and clothing industry is a main constituent in Hong Kong. In this industry, handling a large amount o f images is an important task in various phases, for example, the designing, sourcing, and merchandising phase. We d e v elop an image database system called, Montage for managing and retrieving these visual information e ciently and e ectively. M o n tage is an image database supporting content-based retrieval by color histogram, sketch, texture, a n d shape. One important feature of Montage is the Open Architecture design which makes the system extensible, customizible, and exible. There are two aspects of this open architecture design: (1) Open DataBase Connectivity (ODBC) and (2) plug-in framework which w e will discuss in more details. Moreover, we describe an experimental Java system enabling internet access to Montage. In the paper, we also present a n e x p e r i m e n t t o e v aluate the performance of several query methods.

Content-Based Image Retrieval: A Survey on Local and Global Features Selection, Extraction, Representation, and Evaluation Parameters

“…In visual search, knowledge of the domain of the image plays a vital role as it helps to reduce the gap between the feature description and the semantic description of the image. Though well defined laws are not there for domain knowledge, some papers [4], [5], [6], [7], [8], [9], [10] and [11] have given a few methods to gain knowledge about the domain of the images as summarized below. [5], [6], and [7] discuss the various human perception techniques for describing equality and similarity among images.…”

“…For example, man-made rules can define that images having straight lines and perpendicular corners are of indoor scenes. [11] defines the rules for the fashion, textile, and cloth industries.…”

Visual Search Method for Garment Image Based on SIFT

“…In this paper, we explore the use of computer vision to recognize images of clothing. The key components of the system are the clothes similarity measurement algorithms [5,6,7]. The application described a typical computer vision problem.…”

Distributed content-based visual information retrieval system on peer-to-peer networks

“…Since the mid 1990s, many CBIR systems have been proposed and developed, including QBIC [Faloutsos et al 1994], WebSEEK [Smith and Chang 1997], WBIIS ], SIMPLIcity [Wang et al 2001], MARS [Mehrotra et al 1997], NeTra [Ma and Manjunath 1997], Photobook [Pentland et al 1994], AMORE [Mukherjea et al 1999], Virage [Gupta and Jain 1997], and other systems for domain-specific applications [Lau and King 1998;King and Jin 2001]. The images are preprocessed and are represented in the form of a feature vector with their similarity being based on the distance between the feature vectors.…”