Illuminationnormalized based technique for retinal blood vessel segmentation

The Retinal image carries important information about the health of the sensory part of the visual system. In this paper, a new approach is suggested by utilizing the homomorphic filter integrated with Contrast limited adaptive histogram equalization (CLAHE) method for the illumination normalization and contrast enhancement of the retinal images. Then segmentation is done through several steps by using the existing methods such as morphological filtering, a second derivative operator that is followed by a final morphological filtering stage and hysteresis thresholding. The suggested method is verified on DRIVE and CHASE-DB1 databases and has average accuracy of 72.03% and 64.54%, accordingly. The obtained results demonstrate that the proposed approaches achieve higher accuracy than the traditional method. The suggested approach not only contributes to the successful result, but also minimizes the computing time.

Automated summary

The new method proposed in this paper combines homomorphic filter with CLAHE method for illumination normalization and contrast enhancement of retinal images, achieving higher accuracy. Segmentation of retinal images is done through steps such as morphological filtering, second derivative operator, and hysteresis thresholding, showing positive impact on accuracy improvement and computing time reduction.