International Journal of Innovative Technology and Exploring Engineering (IJITEE) ISSN: 2278-3075, Volume-8 Issue-11, September 2019 1103 Published By: Blue Eyes Intelligence Engineering & Sciences Publication Retrieval Number J11930881019/2019©BEIESP DOI: 10.35940/ijitee.J1193.0981119 ABSTRACT: Content-based image retrieval is a technique which uses visual contents to search images from large scale image databases according to users' interests, has been an active and fast advancing research area since the 1990s. During the past decade, remarkable progress has been made in both theoretical research and system development. However, there remain many challenging research problems that continue to attract researchers from multiple disciplines. Some of the techniques used in CBIR are Query by example, Semantic retrieval, content comparison techniques etc. Most of the existing works were done on spatial domain which is not so efficient. To overcome the difficulties of the existing works, a new algorithm is planned. And the proposed approach is based on the frequency domain for the content based Image retrieval systems.A new image retrieval technique which will retrieve images from image databases based on their contents in frequency domain to get better results. And a relevance feedback method is used for improving the retrieval efficiency. Many techniques are there in this computer vision and image processing field. But using low level features as a basis and retrieve features with good efficiency is problem of the study. The proposed work relates Feature extraction using both frequency domain as well as spatial domain. For spatial domain SIFT and for frequency domain FDCT techniques are applied and results were compared to find better information retrieval. Key wordss: fdct (Fast discrete curvelet Transform), sift (scale invariant function transform), svm (support vector machine ) I. INTRODUCTION Content-based image retrieval (cbir) is the application of laptop imaginative and prescient to the photograph retrieval trouble, i.e. attempting to find digital pictures from huge databases. content material based image retrieval makes use of shade, texture and form features [1] .a machine which can filter out snap shots based totally on their content could offer better indexing and return extra correct effects. in content material based totally commonly image retrieval (cbir),[2] images place unit indexed by way of their visual content like color, texture and form. Those Low-level image alternatives region unit inadequate to explain most internet based generally picture search engines like google agree with strictly on facts and this produces heaps of garbage in the outcomes. Many structures had been developed but now not a unmarried gadget is perfect. green control of the rapidly expanding visual records is wanted. to search the most similar photographs to the question photograph, by way of fast discrete curvelet transforms [3] for better retrieval effects. Image retrieval system may be accomplished both in Frequency domain or spatial area [4]. Inside the frequency area, an photo at every factor represents a selected frequency contained in the spatial area photograph. via applying the changes, on an image it represents within the fourier or frequency domain, at the same time as the input image is the spatial area equivalent. in frequency area wavelet transforms affords a suitable frame work for analysis and characterization of pix at one-of-a-kind scales. wavelet transforms provide a multi-decision approach to texture analysis and category.[5] shengjiu wang (2001) “a sturdy cbir technique the usage of local color histograms” technical document. tr 01-13[6]. j. zhang, g. li, s. he, “texture-primarily based photograph retrieval via facet detection. Relevance comments [7] to modify the retrieval procedure in an effort to generate perceptually and semantically extra meaningful retrieval outcomes. Better retrieval charge may be possessed using relevance remarks [8]. To retrieve an picture with less computational complexity The use of low level functions in frequency domain is predominant motto of the work. Generalizing a cbir gadget is likewise tough as one characteristic will have exclusive significance in specific domain names. The quantity and type of characteristic decided on impacts the output. Complexity using low level features in Frequency Domain is main motto of the work. Generalizing a CBIR system is also difficult as one feature can have different significance in different domains. The number and type of feature selected affects the output. II. CURVELETS A decent frame has been introduced because the curvelet aid to lessen the information redundancy within the frequency domain.curvelets have both variable width and duration and represent more anisotropy. Wrapping based totally curvelet rework is simpler, Less redundant and faster in computation than ridgelet based curvelet transform. Wrapping based curvelet rework is a multi-scale rework with a pyramid shape which includes many orientations at each scale. sub-bands at high and coffee frequency stages have exclusive orientations and positions. at high scales, the curvelet waveform turns into so fine that it looks as if a needle shaped detail .with growth inside the decision degree the curvelet will become finer and smaller inside the spatial domainAnd indicates more sensitivity to curved edges that permits it to successfully capture the curves. curvelets have useful geometric options that set them inside the spacial domain, a curvelet has companion diploma envelope powerfully aligned on a given `ridge' whereas inside the frequency domain; it is supported near a container high frequency components of an CBIR using SIFT& FDCT with Relevance Feedback Mechanism K Sugamya, Suresh Pabboju, A Vinaya Babu