Enriching Image Retrieval System Through CNN for Sketches And Images
Shreya Kukdeja1, A. G. Phakatkar2

1Shreya Kukdeja, Department of Computer Engineering, Pune Institute of Computer Technology, Pune, India.
2Prof. A. G. Phakatkar, Department of Computer Engineering, Pune Institute of Computer Technology, Pune, India.
Manuscript received on 15 August 2019. | Revised Manuscript received on 25 August 2019. | Manuscript published on 30 September 2019. | PP: 1407-1412 | Volume-8 Issue-3 September 2019 | Retrieval Number: B3619078219/19©BEIESP | DOI: 10.35940/ijrte.B3619.098319
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Image retrieval is being a one of the widely research areas in the current paradigm of the software industry. This is mainly due to as a proverb “image can speak many words”, which is because of its diversity of the contained objects and their pattern. Many search engines including Google, is providing the option of searching images by giving query image, most of the time this searching is done by the averaging the features of the images which yields considerable low precision. Convolution neural network and Recurrent neural network are widely used mechanisms to handle the image processing techniques in image retrieval. On using of CNN and RNN many systems are yielding the low accuracy because of the features that they are considering. And again, only finger counting systems are dealing with the image retrieval mechanism for the input of the object rather than the whole image. So as a tiny step towards this, this research article proposes a model of image retrieval using the input as image sketch and images using the histogram features and Region of interest based on the position, volume, color and orientation through the interactive CNN model. The image search is performed using the CNN through K means Clustering and Haar wavelets.|
Index Terms: CNN, Haar Wavelet, Histogram features, Region of Interest.
Scope of the Article:
Image Security