International Journal of Electrical and Computer Engineering (IJECE) Vol. 9, No. 4, August 2019, pp. 2982~2992 ISSN: 2088-8708, DOI: 10.11591/ijece.v9i4.pp2982-2992  2982 Journal homepage: http://iaescore.com/journals/index.php/IJECE Adaptive CSLBP compressed image hashing Varsha Patil, Tanuja Sarode Department of Computer Engineering, TSEC, Mumbai University, India Article Info ABSTRACT Article history: Received Nov 25, 2018 Revised Mar 28, 2019 Accepted Apr 3, 2019 Hashing is popular technique of image authentication to identify malicious attacks and it also allows appearance changes in an image in controlled way. Image hashing is quality summarization of images. Quality summarization implies extraction and representation of powerful low level features in compact form. Proposed adaptive CSLBP compressed hashing method uses modified CSLBP (Center Symmetric Local Binary Pattern) as a basic method for texture extraction and color weight factor derived from L*a*b* color space. Image hash is generated from image texture. Color weight factors are used adaptively in average and difference forms to enhance discrimination capability of hash. For smooth region, averaging of colours used while for non-smooth region, color differencing is used. Adaptive CSLBP histogram is a compressed form of CSLBP and its quality is improved by adaptive color weight factor. Experimental results are demonstrated with two benchmarks, normalized hamming distance and ROC characteristics. Proposed method successfully differentiate between content change and content persevering modifications for color images. Keywords: Authentication CSLBP Histogram Image hashing L*a*b* color model Copyright © 2019 Institute of Advanced Engineering and Science. All rights reserved. Corresponding Author: Varsha Patil, Department of Computer Engineering, TSEC, Mumbai University, India. Email: varshasp2977@gmail.com 1. INTRODUCTION Success and popularity of digital technology is enormous. Digital forgery (tampering) and unauthorized use have reached a significant level that makes multimedia authentication and security very challenging and demanding. Some of this data is confidential and there is need of protecting and verifying the data integrity. It is necessary to protect some data for its confidentiality and integrity. In cryptography, hashing techniques are there for data integrity. These methods are basically designed for text data and follow stringent approach in which even change in single bit drastically causes change in its hash code. Such techniques cannot be utilized for digital data like image, video etc. as limited change is common on these data types. Limited change in the image data indicates content preserving operations like gamma correction, scaling, contrast modification etc. To deal with data integrity issues, image hashing is simple and efficient solution. Content change in an image is treated as malicious operation. The hash code of original and modified image is drastically different or above prescribed threshold when some malicious changes occur in an image [1-4]. Most of the existing image hashing methods target only gray scale images. The proposed hashing method is designed for colour images. For colour image hashing, color is an important feature. However, relying only on colour feature for feature extraction is not sufficient. Texture is a very useful depiction for a wide range of images. Colour is highly correlated, specially RGB colour model whereas structures are uncorrelated and random in nature. Proposed method extracts spatial texture features using modified CSLBP which mainly concentrates on pixel statistics to determine texture strength and pattern. Colour features are fastened in modified CSLBP