A Survey of Content Based Image Retrieval Mechanism

摘要

The conventional CBIR is not perfectly suitable for retrieving the images from the large image dataset. The basic problem is generated from conventional CBIR method are: First, Whenever users perform the image search on google since unnatural and unsupported data is retrieve from www. Second, it is very time consuming, due to this reason perfect result is not obtained in given time duration. Basically Content- based image retrieval is uses for retrieving the similar images from image dataset on basis of sample query image. To overcome the above problem, the proposed scheme improved the performance of image classification and retrieval accuracy of images. The PSO method is used to optimize the feature of image dataset. These feature of reflects texture level of various images. Optimization feature descriptor in the ROI (region of interest) of image using PSO (Particle of Swarm Optimization) Technique. Optimized image descriptors are classified in different classes of image dataset using artificial neural network, ANN is used for image retrieval to maintain accuracy. This supervised learning approach uses the concept of image optimizer to improve the performance of retrieved images. The performance of proposed method ANN-PSO (Artificial Neural Network — Particle Swarm Optimization) method is improved instead of comparable CBIR method.