ANALYSIS OF VARIOUS SEGMENTATION TECHNIQUES FOR OBJECT DETECTION AND RECOGNITION

Abstract

Segmentation technique is used to divide the image into meaningful sub parts. It is also very useful in object recognition and extraction for images or logos. There are various methods of Segmentation among which Threshold method is considered as simplest one. Some techniques are suitable for noisy images. The strongest method of noise cancellation in images is Markov Random Field (MRF). In recent years, a new approach proposed for extraction and recognition of logos in image archives named as Context Dependent Similarity (CDS) kernel. Object Recognition provides domain independent technique for data analysis. Segmentation is very useful in many image applications because it is the first step in image analysis and recognition. The novel variational frameworks are available to match and recognize multiple instances of multiple reference logos in image archives. Test images and Reference logos are seen like constellations of local features and matched by minimizing an energy function mixing like fidelity term and so on. Local features which consider for matching are regions, interest points, etc.