java - ARToolKit: processing of Natural Feature Tracking images (jpegs) on the mobile device itself

Question

I've been using Vuforia for a while now which has the limitation that I can't directly submit to the Natural Feature Tracking processor an image for turning into a trackable data file. Instead it is hard wired to take the image directly from the camera, which gives me no control. See for example the UserDefinedTargets demo.

Does ARToolKit allow me to submit a jpeg to the NFT processor directly from my mobile device for processing ? I want to be able to achieve something like UserDefinedTargets on Vuforia, but with the ability to submit my own natural feature images as jpegs on the mobile device itself. I can then save images taken on the fly for future processing, or even better, save the processed NFT data for future use. I do not want to use some cloud service, e.g. there is a workaround with Vuforia, but I have to use their cloud service and that has its limitations too !

Answer 1

根据此处的文档：http://artoolkit.org/documentation/doku.php?id=3_Marker_Training: marker_nft_training您有一个可用于进行特征提取的程序。它适用于数字图像，因此无需查看代码，我为您预测了两个选项：

a) 查看源代码，看看您是否可以让该工具在 Android 手机中运行，很可能是通过 NDK

b) 制作一个接收图像、运行该程序并返回结果的 Web 服务，因此您可以将其用作普通的 REST API。

希望这有帮助。