Heuristic Algorithm to Generate Modified Freeman Chain Code from Thinned Binary Image

Chain code is an image representation which can be used to represent a shape of object or structure and also to represent connectivity between lines in the image boundary. It can be used in various applications because of its ability for information preservation and allows considerable storage space reduction for properties data image shape. This representation also can be applied in image processing field such as image compression, feature extraction and pattern recognition. Extracting chain code for boundary image or shape of object is simpler compared to extracting two-dimensional thinned binary image (TBI) that contain junctions. Thus, this paper presents a new chain code scheme and its algorithm to extract the chain code from TBI with multiple junctions. The importance of this chain code is mainly for feature extraction and recognition processes against such images. Before extracting the chain code, TBI with location-marked junctions is required as input data. This input data is a text file contains thinned binary image (0,1) plus junction marker, ‘J’ character to indicate a corner or junction at corresponding location. Junction positioning and labelling can be performed manually or by using corner/junction detection algorithm. Subsequently, the input file (0,1,J) will be traversed starting from image boundary and is followed by its inner line. In traversing process, all junction markers will be sequentially renamed to character A-Z to distinguish among existing junctions and, MFCC will be generated simultaneously. The peculiar way this MFCC is generated is due to feature extraction and recognition needs.