Showing posts from October, 2014

iBeacon heat map test

Generating heat map using iBeacons with android application Intro About a month ago I had an opportunity to play with an iBeacon on the Android platform. My first task was to implement a proximity trigger to Android client for one of our apps at work. As we played with it, we wanted a more precise data such as a heat map. Searching I googled some familiar projects around, bud didn't find a precise app in the Play Store that generates a heat map from actual distances from iBeacons. After visualization the raw accuracies in a graph (picture below) I found this interesting page that nevertheless confirm my results. These are my observations. raw data is not as accurate as I first thought accuracy depends on rotation of iBeacon to the smart phone at some distance, the error is rising rapidly to get more precise data, you have to set a transmit interval and transmit power to the maximum, but that also leads to more battery consumption What I did I created an Android

Bill cropper

Automatic tool for cropping bills from scan What is it? We have this project at work, that is intended for small and medium companies. It covers much more than just simple accounting. It has many cool features that I'm not going to write about, except one. Automatic processing bills/ invoices. You can do it via your mobile client (Android, iOS) or a web application. What to improve? We ran into a problem that people scanned more bills together, instead of one image - one document rule. So I decided to create a small script to recognize and crop these multi-bills images into separate images. Algorithm To crop the bills out of the images I used a few simple preprocessing image methods. First, I remove the noise. After that I convert the image to binary and separate it into white areas. Finally I find enough big white areas and find a bounding box for them. I think the code below is self explaining enough. 1. RESIZE 2. IMOPEN - MORPHOLOGICAL OPERATION 3. IMG TO BINAR