Artificial Neural Networks for Image Understanding
Artificial Neural Networks for Image Understanding
| Hi-index | 0.00 |
Biologists data are often in the form of images. Much information can be extracted from such images. To extract the necessary information, biologist have to spent long and labourious hours studying the images. Computers can now be trained or programmed to perform such task. One way to achieve this is to apply Artifical Neural Networks (ANN). However, before ANN may be applied to the digitized images must be pre-processed to ensure the images are relatively clean. In this research, we investigate the feasibility of classifying lichen images into the different species using ANN. The process involves the extraction of classification features from the images for the input to a template matching technique employing an ANN algorithm. The images studied are pre-processed to improve its quality prior to the feature extraction phase. The pre-processing stage deals with problems related to variable shape sizes, colour, background noise and variable placement angles of the shapes. With the implementation of the above techniques, we are able to achieve an accuracy more than 90% for the classification the lichen images into their species.