Several countries around the world use CCTV systems as forensic evidence to combat crime. These cameras cover large fields of view, where low-resolution facial images are typically captured, making the identification of the subject of interest very difficult. Distortions caused by video compression, motion blur, and poor lighting conditions can also further reduce quality and thus reducing their effectiveness. These methods are in most cases insufficient, especially when dealing with dynamic non-rigid objects such as faces.
DEEP FIR aims to improve the quality of facial images captured by CCTV cameras using models optimized to restore compressed low-resolution facial images typically found in CCTV footages through an advanced artificial intelligence (AI) technique based on deep-learning to restore very low-resolution and compressed images. The developed algorithm has shown that it is capable of restoring degradations that are typically present in CCTV footage.
The figure shows a number of low-quality facial images on the left and the restored images on the right. By comparing the restored face to the actual high-resolution face, one can see that the proposed method is able to significantly improve the quality of the face while preserving the identity of the person of interest.
The technology aids image restoration which is applicable for law enforcement, forensic laboratories and security agencies/ companies.
This approach was tested in a controlled environment and has shown promising results. The team is now working on improving the performance of the existing algorithm and extending it to generalize to the different degradations produced by different CCTV cameras. The correlation present in different video frames will be exploited to develop a multi-frame face super-resolution algorithm.
Dr Ing. Reuben Farrugia
Access the DEEP-FIR Project website
Financed by the Malta Council for Science & Technology (MCST), for and on behalf of the Foundation for Science and Technology, through the FUSION: R&I Technology Development Programme.