THEORETICAL ADVANCES Enhancing fractal image compression speed using local features for reducing search space Keyvan Jaferzadeh 1 • Inkyu Moon 1 • Samaneh Gholami 1 Received: 3 January 2015 / Accepted: 4 May 2016 Ó Springer-Verlag London 2016 Abstract The encoding step in full-search fractal image compression is time intensive because a sequential search through a massive domain pool has to be executed to find the best-matched domain for every range block. To afford a fair encoding time, immaterial domain–range block comparisons should be prevented. In this paper, a new local binary feature resemble to local binary patterns method is introduced. This single local feature is robust to noise and can exploit the general structure of the block. Concerning similarity between range–domain blocks, a criterion is allocated dynamically by measuring the pixel diversity among the range block pixels. To avoid redun- dant calculations, the distance of the general pattern is assessed by the Hamming distance utilizing a pre-com- puted table. Experimental results show that the presented approach can make FIC a lot faster as opposed to the full- search method and outperform some other identical methods while preserving the quality of the decoded images. Indeed, the proposed method can be utilized inside identical applications that want a specific block size or blocks comparing. Keywords Fractal image compression  Local features  Hamming distance  Local binary pattern  Adaptive thresholding 1 Introduction The reduction of data storage requirements is the main outcome of data compression methods. Furthermore, compression offers an attractive approach to decrease the communication cost while transferring substantial volumes of data through links via a relatively high effective uti- lization of the bandwidth of the available links. Due to the reduction in the data rate, the cost of communication can significantly make use of these methods. The fractal image compression (FIC) algorithm has received considerable attention not only as a powerful image compression technique [1] but also for applications to dif- ferent areas such as image restoration, medical image clas- sification, watermarking, shape recognition, and face recognition [2–6]. Furthermore, the decoding of fractal encoded images is straightforward, fast, and very easy to implement. Besides, the resolution-free decoding property is another advantage of FIC, which means that the decoder can retrieve compressed images in different zooming scales. Microsoft Multimedia Encyclopedia (Microsoft Encarta), for example, used FIC to compress a large number of images, which practically shows that the properties of FIC make it suitable for multimedia applications [1]. The original idea of FIC was proposed by Barnsley [7] in which a considerable amount of redundancy existing in the images could be explored. To eliminate redundancies, Jac- quin [8] proposed a partitioned iterated function system (PIFS) that contains contractive transformations for each image that together have a fixed point similar to that of the original image. This implies that by applying the transforms iteratively on an arbitrary initial image, the output image will converge to a fixed point similar to that of the original image. As such, storing these transformations means less space, and this can be considered an effective compression method [9]. & Keyvan Jaferzadeh kjaferzadeh@chosun.kr 1 Department of Computer Engineering, Chosun University, 309 Pilmun-daero, Dong-gu, Gwangju 501-759, South Korea 123 Pattern Anal Applic DOI 10.1007/s10044-016-0551-1