For the final project, start with an application or problem that you want to solve, then identify appropriate DSP algorithms and implementation methods.
You are strongly encouraged to develop your own project idea and identify an appropriate DSP algorithm. If you choose one of the example topics below, you should still investigate possible algorithms and make your own design choices rather than simply reproducing a previous implementation.
A good project should have:
a clearly defined input and output;
a meaningful signal-processing component;
a reasonably constrained working scenario;
a quantitative or otherwise well-defined way to evaluate whether the system works; and
an algorithm that your group can understand and implement.
A useful way to develop your project idea is to answer the following questions:
What problem do you want to solve?
Example: recognize which exercise a user is performing.
What signal can you measure?
Example: accelerometer and gyroscope measurements.
What information do you want to extract?
Example: exercise type and repetition count.
What DSP operations or algorithms could extract that information?
Example: filtering, segmentation, feature extraction, correlation, spectral analysis, or classification.
What assumptions will make the problem manageable?
Example: only four predefined exercises are considered, and the tablet is attached at a fixed location.
How will you test whether the system works?
Example: measure classification accuracy over multiple users and trials.
Use accelerometer, gyroscope, or other IMU measurements to estimate motion or recognize activities.
Example directions:
exercise recognition or repetition counting
gesture recognition
posture or motion classification
step or gait analysis
simple indoor motion tracking
sports-motion analysis
Use microphone or audio signals to detect, classify, estimate, or modify acoustic information.
Example directions:
song or music identification
speaker recognition
sound-event classification
tempo or beat estimation
acoustic ranging
echo or delay estimation
noise reduction
speech or music analysis
equalization
Use camera images to extract information, detect objects or structures, or recognize patterns.
Example directions:
document or ID-card scanning
edge- or line-based object detection
barcode or QR-related processing
digit or symbol recognition
shape detection
image enhancement
perspective correction
image-based measurement
Process sequences of images to estimate motion or extract information that cannot be obtained from a single frame.
Example directions:
video stabilization
object tracking
motion classification
trajectory estimation
camera-motion estimation
temporal filtering