Turkish Journal of Electrical Engineering and Computer Sciences




Activated sludge samples were taken from the Konya Wastewater Treatment Plant. Two hundred images for each sample were captured by a systematic examination of the slides. Segmentation of microscopic images is a challenging process due to lack of focus. Therefore, adjustment of the focus is required for every movement of the mobile stage. Because the mobile stage does not have the z axis, the focus cannot be adjusted. A new method that uses automatic segmentation of the captured images is developed for solving this problem. The proposed method is not dependent on image content, has minimal computation complexity, and is robust to noise. This method uses a cellular neural network (CNN) in which an adaptive iterative value is calculated by wavelet transform and spatial frequency. A model is fixed in the system in order to estimate the iterative value of the CNN. Integrated automatic image capture and automatic analysis of large numbers of images by using evaluation software are improved in our system. Approximately 1000 microscopic images are processed in this experiment. The proposed method is compared with the traditional threshold method and the CNN through constant iteration. The experimental results are given. \vs{-1mm}


Automatic image capture, wastewater treatment, segmentation, activated sludge, cellular neural network, wavelet transform, spatial frequency, entropy

First Page


Last Page