| Summary: | In this study, we present a novel and automated system for estimating human body’s pH fluids from non-calibrated camera phone images. Two approaches have been devised and their performance has also been evaluated. First approach is based on discontinuity segmentation of images using edge detection filters and second approach employs color quantization techniques to produce classification results. We have also implemented two J2ME-based prototypes to test and evaluate the performances of two proposed methods on three different mobile camera phone devices with various image resolutions. Experimental results showed that pH levels reported by the two programs are consistent with the ground truth estimation. Compared to certain existing solutions like personal pH meters, our software solution is more affordable, requires no or very little hardware infrastructure to operate, and can also be upgraded easily. In the long term, our goal is to develop more complex mobile medical solutions for personal healthcare purposes such as urinalysis by combining computer vision and mobile computing disciplines.
|
|---|