Construction of garment pattern shape information system using image analysis and shape recognition techniques

Abstract

Purpose - As an extended work of the previous paper (Sul, 2010), this paper provides a guideline information for an anonymous garment pattern in sewing process. The purpose of this paper is to first, provide garment pattern database. By simply taking pictures of garment patterns, the shape database is constructed. Once the shape database is prepared, data retrieval can be done by image indexing, i.e., simply inserting garment pattern boundary shape again to the database. Using shock graph methodology, the pattern sets used for database preparation can be exactly retrieved. Second, to find the nearest shape of a given input pattern shape in the database. If the input garment pattern shape does not exist in the database, the shape matching algorithm provides the next similar pattern data. The user, who is assumed to be non-expert in garment sewing process, can easily predict the position and combination information of various patterns. Design/methodology/approach - Image processing is used to construct the garment pattern shape database. The boundary shapes are extracted from the photographs of garment patterns and their shape recognition information, especially shock graph, is also recorded for later pattern data retrieval. Findings - Using the image processing technique, garment patterns can be converted to electronic format easily. Also the prepared pattern database can be used for finding the nearest shape of an additional given input garment pattern. Patterns with irregular shapes were retrieved easily, while those with a simple shape, such as rectangle, showed a little erroneous result. Originality/value - Shape recognition has been adopted in various industrial areas, except for garment sewing process. Using the provided methodology, garment pattern shapes can be easily saved and retrieved only by taking pictures of them.