2012 INTERNATIONAL CONFERENCE OF KIICE(ICKIICE 2012) 1 Abstract—We design a new footprint detection program for implementing a biological database and for footprint recognition in order to support biosecurity especially for environmental surveillance. For this programl design, we propose a footprint detection and a feature extraction solution to solve several main issues such as incomplete, fading, missing, sliding, overlapped, or asymmetrical footprints. Extracted footprint images and features are used to develop a new online footprint database for better productivity, reliability, portability and source management for environmental surveillance studies (e.g. invasive species eradication programs). Index Terms—Environmental surveillance, Footprint Recognition, Image processing, Database, Feature Extraction I. INTRODUCTION Footprint recognition has always been one of the most efficient and economical solutions to identify and track species. Ecology experts use a standard tracking tunnel system to identify and track small species. As the tracking tunnel system requires completely no electronic power [1], it is a very economical and effective way to estimate the current animal distribution as most of the areas are not fully accessible. Unfortunately, only the very experienced biologists are able to identify the type of species from tracking cards. Even if the biologists are willing and able to identify the tracking cards manually, it will take them an amount of time. Human evaluations also vary, and a program stays to its rules. These are two of the reasons why we need to build a footprint recognition database. Qualities of collected footprint images are normally very poor. This makes the whole footprint recognition process extremely inefficient and unreliable [2]. Therefore, we propose to design a novel footprint detection program for an online database for footprint recognition. This new program is not a just for trans- ferring image cards into image data. The core ideas of the program are to integrate image processing abilities that can automatically improve the quality and reliability of a scanned image before it is subjected to further analysis. II. TRACK ACQUISITION The footprint cards are all collected by a tracking tun- nel system consisting of tunnel, lures and a pre-inked tracking card. The lures attract small species such as rats or mice to walk through the tunnel. As the tracking card is pre-inked, the species leave their footprints. The tunnel is just simply designed for protecting the tracking card from rain or other animals such as bird [3]. The tracking tunnel system is shown below in Figure 1. Fig.1. Tracking Tunnel System [4] III. FOOTPRINT DETECTION MODEL Online biological databases have been developed and published. For example, the Global Invasive Species Database (GISD) is one of them that is developed by The University of Auckland, managed by the Invasive Species Specialist Group (ISSG) [5]. The GISD provides free resources of information on invasive species and eradication details. It has increased the public awareness about invasive species and also gives a standard guide of invasive species eradications. A. Challenges and Issues for Footprint Detection Before we transfer the images from tracking cards to the database, we have to solve several issues. Basically, we identified five main issues for the footprint images: incomplete, fading, missing, sliding, or overlapped footprints. 1) Incomplete footprints Not all areas of toes contact the tracking card. This is where incomplete footprints occur as shown below in Design of a Footprint Detection Program for a Biological Database Jerry Zheng, Bok-Suk Shin and Reinhard Klette __________ All authors are with the University of Auckland, New Zealand.