A Real Time Hough Transform Architecture Useable inside a WCE

Une architecture de transformation difficile en temps réel utilisable dans un WCE

Orlando Chuquimia (1) , Andrea Pinna (1) , Xavier Dray (2) , Bertrand Granado (1)

1 SYEL - Systèmes Electroniques
2 CHU Saint-Antoine [AP-HP]

To reduce the incidence of colorectal cancer (CRC), we propose a new paradigm of Wireless Capsule Endoscopy (WCE) that recognizes polyps in-situ. We embed an image processing chain in a System on Chip (SoC) that uses the Hough Transform as part of the processing to detect circles in High Definition (HD, 1920x1080) images. A circle here is a probable marker of colorectal cancer ; a polyp. Hough Transform is a widely used shape-based algorithm for object detection and localization. This technique can be generalized to circles. To embed the Hough Transform inside a WCE, considering real time execution and a limited area, several optimizations are necessary due to computational requirements. This paper presents an efficient, real-time architecture that can be integrated in an 8x8mm 2 embedded system of a HT algorithm for multi-circle detection. The architecture's area has been validated in a FPGA Xilinx Spartan 7 XC7S15-CPGA196 packaged in an area of 8x8mm 2. This architecture can run at 135.46MHz on a 1920x1080 pixels image.

Domaines: Architectures Matérielles [cs.AR]; Intelligence artificielle [cs.AI]; Systèmes embarqués; Imagerie médicale; Traitement des images [eess.IV]

Orlando Chuquimia, Andrea Pinna, Xavier Dray, Bertrand Granado. A Real Time Hough Transform Architecture Useable inside a WCE. 2019 IEEE Biomedical Circuits and Systems Conference (BioCAS), Oct 2019, Nara, Japan. pp.1-4, ⟨10.1109/BIOCAS.2019.8919052⟩. ⟨hal-02495530⟩