A REVIEW ON OBJECT IDENTIFICATION

Abstract

Object identification is an active research area with numerous commercial applications. Image of object is represented as single or multiple arrays of pixel values. Features that uniquely characterize the object are determined. The arrays are compared with a stored pattern feature set obtained during training procedure. Number of matches of the object in the image must be obtained. As the image consists of a large amount of data, it has to be compressed using a compression technique so that data reduction is achieved. This reduced data is used for comparison process. This image identification technique can be used to recognize objects in specific areas. In this specific work compressive sensing using Sparse Representation can be used for object identification. The planned object identification system consists of three main stages, Sparse dictionary design, Feature Extraction using down sampling and Object detection utilizing sparse representation. Initially a sparse dictionary will likely be designed along with trained with large variety of different training images. It images incorporate assorted items including target image. Intended for feature extraction, if sparsity inside the recognition issue is correctly harnessed, the choice of features is not a critical issue. However, the leading problem can be whether the volume of features can be sufficiently substantial and whether the sparse representation is the right way computed. Here down sampling process inside the images will likely be done. In assessment process, the sparse dictionary typically determines the object among the other items. Extensive experiments will likely be conducted to help verify the efficacy in the proposed criteria, and corroborate the above-mentioned claims.