hybrid learning method

 hybrid learning method

compared with the conventional learning method for solving a real-world problem of image enhancement. Then, a simple, fast, and efficient algorithm is designed to solve the problem. Moreover, the proposed method is suitable for the application of image processing on a mobile device. We apply our method on a mobile device by optimizing the algorithm for the performance on low memory condition and power consumption. Experimental results show that our proposed method achieves the higher accuracy with less number of the parameters. In the future, we will make full use of more complex model to further increase the accuracy. .M., H.C. and J.S. conceived the idea. C.M. designed and implemented the algorithm. C.M. and H.C. conducted the experiments. C.M., H.C. and J.S. analysed the results. J.S. wrote the manuscript. All authors reviewed the manuscript. This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and future Planning (No. 2015R1D1A3A01019588) and by the Institute for Information & Communications Technology Promotion(IITP) grant funded by the Korean government(MSIP) (No. 2017-0-01798, No. 2017-0-01990, No. 2018-0-00983, and No. 2018-0-00993). The authors declare no conflict of interest. [Example of image enhancement. (**a**) Original image; (**b**) blurred image; (**c**) the output image by the conventional approach; (**d**) the output image by our approach.](sensors-18-02497-g001){#sensors-18-02497-f001} [An example of a convolutional neural network.](sensors-18-02497-g002){#sensors-18-02497-f002} [An example of a hybrid learning method: (**a**) A simple case of hybrid learning; (**b**) the procedure of the hybrid learning.](sensors-18-02497-g003){#sensors-18-02497-f003} [The overall structure of the proposed method.](sensors-18-02497-g004){#sensors-18-02497-f004} [The structure of our enhancement module.](sensors-18-02497-g005){#sensors-18-02497-f005} [The architecture of the proposed end-to-end network for the enhancement module.](sensors-18-02497-g006){#sensors-18-02497-f006} [Comparison of images that we have enhanced by the proposed algorithm: (**a**) Original image; (**b**) The image by the conventional method; (**c**) The image by our proposed method.](sensors-18-02497-g007){#sensors-18-02497-f007} Comparison of the accuracy of

https://claraeon.com/