IMAGE PROCESSING FOR DETECTION OF PRODUCT TAMPERING USING A CAMERA

Vo Nguyen Khang
April 2, 2012

Abstract

PIN entry device (PED) is one type of devices that require high security protection since it directly receives information about customer’s card and PIN. Many methods has been applied to make the PEDs tamper-proof, including complicated sensing circuits, but PEDs still remains vulnerable to many attacks, typically in the type of object intrusion. Anti-tampering is also the primary concern in many other systems, such as voting machine or other devices in the defense industry.

This project will explore another tamper-resistance method using computer vision technique to protect the product at the PCB assembly level. Every time the device is powered on, the image of the PCB is captured and compared with the reference image to check whether suspicious components have been inserted into the device. The approach used in this project is background subtraction with image registration as the pre-processing step to enhance accuracy. With this approach, most of the test objects are detected successfully, even for small objects such as a 1.25 × 2 mm SMD capacitor or even a 0.25mm diameter AWG30 wire. Although there is much work to be done before this approach becomes available on real devices, this project successfully demonstrates the idea of using computer vision in this anti-tampering application.

Links:

This project is initiated by VeriFone Singapore Pte Ltd under the industry collaboration scheme.