Supplemental Material for A Spectral Information Coding System for the Visually Impaired

MOVIE 1 This movie dipicts that the proposed system can detect the target in real time. The sparkling red LED lights were connected with micro vibrator motors; besides, each motor and corresponding LED were turned on or off at the same time. ( MOVIE 1 is uploaded at https://youtu.be/f9H0wVne7VI ) MOVIE 2 This movie depicts that the user can locate stationary target in real time. Each sparking red LED indicates the status of the corresponding vibrator motor. ( MOVIE 2 is uploaded to https://youtu.be/8yBgNBnepwg ) MOVIE 3 This movie describes how the real-time object tracking works. The right screen displays input image, and the left screen displays the processed image. The purple symbol of zero was selected as target, and the software filtered out other colors in the input image. The left screen only displays the regions of the selected color. The regions are featured by an artificial neural network, and mark the target with white square and others with red square. The result shows that the artificial intelligence can differ the target (purple symbol of zero) from others (purple scarf) ( MOVIE 3 is uploaded to https://youtu.be/Z4Bq4LYqJwA )

影片1：本影片展示了所提出的系统可实现目标的实时检测。闪亮的红色发光二极管（LED）与微型振动电机相连；此外，每个电机与对应的LED会同步启闭。（影片1上传地址：https://youtu.be/f9H0wVne7VI）
影片2：本影片展示了用户可实现静止目标的实时定位。每个闪烁的红色LED用于指示对应振动电机的工作状态。（影片2上传地址：https://youtu.be/8yBgNBnepwg）
影片3：本影片演示了实时目标跟踪的工作原理。右侧屏幕展示原始输入图像，左侧屏幕展示经处理后的图像。本次选取紫色零符号作为跟踪目标，软件会对输入图像中的其余颜色进行滤除，左侧屏幕仅保留所选颜色对应的区域。该区域经人工神经网络（artificial neural network）提取特征后，由软件以白色方框标注目标区域，其余区域以红色方框标注。实验结果表明，本人工智能系统可将目标（紫色零符号）与其他干扰物（紫色围巾）有效区分。（影片3上传地址：https://youtu.be/Z4Bq4LYqJwA）