Image processing strategy used for simulated phosphene map of artificial vision

Longmin Chen, Fanghua Yan, Zezhi Gong, Yayu Zheng

COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12B) 698-703

College of Information Engineering, Zhejiang University of Technology, No.288 Liuhe Road, Hangzhou, China

Phosphene generating mechanism is the most important theoretical foundation and key technology for artificial vision. Simulating the mapping relationship between visual image information and limited phosphene maps remains a difficult problem. To satisfy the practical requirement of clinical trials, we present an image processing strategy for simulated phosphene map. Based on block segmentation, a reducing pixelated image processing method is proposed. Meanwhile, an electrode intensity control strategy based on brightness grading is carried out as well. Finally spatial response experiment is performed on the artificial vision platform based on DSP to prove the algorithm availability. The experiment indicates that under the premise of ensuring the electrode stimulation accuracy, this system can precisely extract the brightness information of the pixel block and the contour information and transform them into electrode stimulus. The speed of the image processing strategy based on DSP is up to 30 frames per second after being optimized, which meets the real time needs of visual centers system completely.