Distributed canny edge detection algorithm using morphological filter

Morphological Filter based Distributed Canny edge detection algorithm for Raspberry Pi platform using Simulink model is presented in this paper. Traditional canny edge detection algorithm uses frame based statistics which gives high accuracy but computationally more complex. Also canny algorithm is more sensitive to noise. In this experiment, an attempt is made to make canny algorithm more robust to noise using morphological filtering. Here canny algorithm is implemented at block level without any compromise in edge detection performance. If frame level statistics are used for threshold selection, it would result in either loss of edges or surplus edge detection. To solve this problem threshold selection is made based on type of block. Smooth and texture pixel counts are calculated for image block. Instead of using probability, actual pixel counts are used to calculate threshold. This makes threshold selection block more adaptive. Finally, objective analysis is carried out which shows proposed block based distributed algorithm is better than traditional frame based algorithm, especially in presence of impulse noise.