2 of 3

Video Capture (Webcam)

Embedded Linux Machine

(eg. Jetson)

MJPEG Stream

(through Wi-Fi)

Unity App

(downloaded to Hololens)

Hololens

Grayscale

(Removes color noise)

Histogram Equalization

(Balances contrast)

Gaussian Blur

(Removes noise grain)

Laplacian Sharpening

(Restores sharp edges)

Linear Brightness

(+25% gain)

Key References

Buades, A., Coll, B., & Morel, J.-M. (2005). A non-local algorithm for image denoising. In 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05) (Vol. 2, pp. 60–65). IEEE.
Land, E. H., & McCann, J. J. (1971). Lightness and retinex theory. Journal of the Optical Society of America, 61(1), 1–11.
Pizer, S. M., Amburn, E. P., Austin, J. D., Cromartie, R., Geselowitz, A., Greer, T., ter Haar Romeny, B., Zimmerman, J. B., & Zuiderveld, K. (1987). Adaptive histogram equalization and its variations. Computer Vision, Graphics, and Image Processing, 39(3), 355–368.

Conclusion

Example applications in Asia:

warehouse and factory robots working in dim environments
low-light industrial inspection in tunnels, plants, and infrastructure
wearable situational awareness for field workers and emergency response
affordable perception systems for student labs, startups, and regional deployment

Fig. 6 AR &VR Application

Supports Asia’s technological future through low-cost, deployable innovation, not expensive or compute-heavy solutions. This matches the conference’s focus on research shaping Asia’s social, economic, and technological transformation.

Streaming

LOW-COST NIGHT VISION CAMERA BASED ON A MODIFIED CON-SUMER WEBCAM AND LIGHTWEIGHT REAL-TIME ENHANCEMENT

Research question

How can low-cost real-time night vision be achieved?

Brief Context

Many night-vision systems are expensive or power-hungry
Low-light enhancement often increases noise and blur
Stronger methods improve visibility but reduce real-time speed
Deep-learning solutions need more compute and training data
Few studies compare low-cost hardware + lightweight enhancement under the same controlled conditions
Need a solution that balances cost, clarity, and real-time performance

Authors: Yevgeniy Dikun, Zaki Al-Farabi, Mirat Serik | Nazarbayev University, ISSAI, Tactile Lab, ARMS Lab

Methods

Key findings

Hardware modifications

The mechanical removal of IR-cut filter from a camera is the most essential step. Without this hardware modification camera will not be able to obtain sufficient data for further modifications.

Effectiveness of processing

Grayscale → CLAHE → Denoise → Sharpen - such sequential approach is superior to using single complex algorithm. Each step does specific enhancements and successfully preserves details and reduces artefacts.

Latency

The lightweight processing pipeline introduces small latency compared to many other methods. This allows to transmit image to HoloLens headset with low latency and use it.

Additional key findings

Retinex methods → higher quality but slow (12–23 FPS)

NLM denoising → very slow (~4 FPS)

Thermal camera → best visibility but high cost

Event camera → low latency but needs dynamic illumination

Main bottleneck: network streaming, not processing

Image Processing

Fig. 1 Setup of our Solution

Fig. 3 Software Comparisons.

Fig. 2 Demonstration of filters’ effect as seen figure 5

Fig. 5 Flow Diagram

Figure 4. Hardware Comparisons.

Physics

Congress

Nazarbayev University 2026

International Symposium on Optical Sensors 2026

1 of 3

2 of 3

3 of 3