A framework for fast automatic image cropping based on deep saliency map detection and gaussian filter

ABSTRACT

Extracting robust visual saliency map and image cropping are fundamental problems in computer vision, graphics, and so on. It is not easy task to accurately detect and crop the entire salient object from images with complex background. In this paper, a deep learning strategy is adopted to train a large data-set of images, to get saliency map from the input image using graph-based segmentation and gray level adjustment to enhance and extract more accurate and clear saliency map. Furthermore, the Gaussian filter and image scaling used along with cropping method to keep better presentation of the visual object. The important task of overall framework is to take care about relevant image contents as well as to identify more region of interest and get optimum rectangle from the saliency map with minimum and maximum rectangular windows. Quality and low computational complexity have been focused while performing the cropping operation because automatic and efficient cropping technique should not only rely on geometric constraints, but it should be fast enough to consider the important image contents. The applied method use different data-set of images, to ensure the efficiency of this technique and the experimental results show that the framework is not only fast as well as much better for image cropping. We used Matlab and Caffe framework for efficient experimental results.

KEYWORDS:

• Automatic Image Cropping
• saliency map
• salient object detection
• visual attention

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Ziaur Rahman received MS degree in software engineering in 2017 from Chongqing University, Chongqing, China. Currently, He is pursuing PhD degree from Sichuan University, Chengdu, China. His research includes are image processing, deep learning and fractional calculus.

Yi-Fei Pu received the PhD degree from the College of Electronics and Information Engineering, Sichuan University, in 2006. He is currently a full professor and a doctoral supervisor with the College of Computer Science, Sichuan University, a chief technology officer with Chengdu PU Chip Science and Technology Company, Ltd., and is elected into the Thousand Talents Program of Sichuan Province and the Academic and Technical Leader of Sichuan Province. He has first authored about 20 papers indexed by SCI in journals, such as the International Journal of Neural Systems, the IEEE TRANSACTIONS ON IMAGE PROCESSING, the IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, the IEEE ACCESS, the Mathematical Methods in Applied Sciences, the Science in China Series F: Information Sciences, and the Science China Information Sciences. He held several research projects, such as the National Nature Science foundation of China and the Returned Overseas Chinese Scholars Project of Education Ministry of China., and holds 13 China Inventive Patents, as the first or single inventor. He focuses on the application of fractional calculus and fractional partial differential equation to signal analysis, signal processing, image processing, circuits and systems, and machine intelligence.

Muhammad Aamir was born in (1986). He received his Bachelor of Engineering Degree in Computer Systems Engineering from the Mehran University of Engineering & Technology Jamshoro, Sindh, Pakistan in (2008) and Master of Engineering Degree in Software Engineering from CHONGQING University P.R. China in (2014). Currently, he is a PhD research student in Sichuan University P.R. China. His research interest includes Data Mining repositories, Image processing, deep learning and fractional calculus.

Farhan Ullah received MS degree in 2012 from CECOS University Peshawar, Pakistan. Currently, he is on PhD Study leave from lecturer position CIIT Sahiwal, Pakistan. He is pursuing his PhD degree in 2017 from Sichuan University, Chengdu, China. His research interests are Software Similarity, Data Science and IoT.

Additional information

Funding

The work was supported by the National Natural Science Foundation of China [grant number 61571312]; Academic and Technical Leaders Support Foundation of Sichuan province [grant number (2016)183-5]; National Key Research and Development Program Foundation of China [grant number 2017YFB0802300].