The need to detect and localize activity exists in a variety of fields and has broad applications ranging from brain activity for neuroscience to image/video change detection for scene surveillance. In this paper, we introduce a method for localizing activity in integrated circuits (IC) using a laser scanning microscope. Specifically, we develop an activity measure using the Generalized Likelihood Ratio Test (GLRT) framework to detect if an electronic waveform is present at a laser scan location while a test program is repeatedly run. By plotting the activity score against laser scan location, an activity image is developed showing the measured activity in different parts of the IC. We show that this technique allows detection of all active regions in an operating IC, while requiring a relatively low number of signal acquisitions.

We systematically compared the optical performance of two computational photography (CP) approaches used to extend the depth of field of an imaging system: wavefront coding and lattice focal lens imaging. A test bed was developed by translating test resolution targets through the focal plane of a high magnification telescope placed 25 meters away. We demonstrated an increase in depth of field from 2 cm to 20 cm in each case. We found qualitative differences in restored image quality that depend both on the CP modality and signal to noise. A comparison of results in the frequency domain suggests optimization strategies for computational photography when the frequency content of imaged objects is known ahead of time.

We present a non-invasive method for recognition of components in a digital CMOS integrated circuit (IC). We use a confocal infrared laser scanning optical microscope to collect multimodal images through the backside of the IC. Individual modes correspond to passive reflectivity measurements or active measurements, such as light-induced voltage alteration. The modes are registered and stored in a multidimensional data cube. We apply a machine learning algorithm using a binary representation to identify a variety of data structures from transistors to entire logic cells. Because of the compact representation, objects can be detected rapidly. We show that by increasing the number of imaging modes used to develop the descriptor, we can significantly increase recognition accuracy. The approach allows recognition of poorly resolved components, whose primary distinguishing features are below traditional optical resolution limits, and is general enough to be applied to multiple design processes. We believe this represents a significant step toward a fully non-invasive IC reverse engineering system.