Search

In this study we propose a system that uses image processing techniques to predict the growth rate, fetal head recognition and head length from fetal image. In the processing of images, we use various methods to improve the image by pre-processing and to try to extract other features from the image. Pre-processing, segmentation, and extraction of features are the three general steps that all types of data will undergo while using the technique. This work focuses primarily on developing a system for predicting whether fetal growth is normal or abnormal and extracting features such as fetal duration and BPD (Biparieatal Diameter). We are doing the research trimester-wise processing of ultrasound images from various pregnant ladies in our proposed job. Using wiener filter, the ultrasound images are smoothed and the image is segmented by border cleaning thresholding, and morphological operations. [ABSTRACT FROM AUTHOR]

The initial stage of the document image is pre-processing. When we see historical document it will be degraded and it need to be preserved. Here in this paper we are considering the aged historical document of agreement copy and we are performing binarization to the document. Many methods has been proposed for binarization but there no perfect method for each every type of the document. So we propose a novel method by using both local thresholding and global thresholding and check their accuracy of the each method. We use Otsu, Niblack, sauvola, wolf methods and added morphological operation like dilation and erosion which gives extra clarity to the document. The dataset we have collect is from both Kannada and Malayalam. Promising result was seen with the accuracy of 87.03% which is better than other proposed methods for the agreement copy. [ABSTRACT FROM AUTHOR]