Figure 4-1 shows the proposed design.
The proposed design has been divided into two parts. In the first part, two modules of the project are shown. The first module is database creation which involves feature extraction and storing the values of all the features in a file called “features.dat” and storing the target vectors of all images in the file “outType.dat”. The feature values and target vectors, which are extracted in first module are used in training the neural network which is second module of the first part. Only one time execution is sufficient for first part, subsequent executions are necessary only if the files “features.dat” (containing shape feature values); “outType.
- Thesis Statement
- Structure and Outline
- Voice and Grammar
dat” (containing target vectors) and “net_FFBP.mat” (containing the trained information of Neural Network) are not present in the system in which the project is being executed.In the second part there are two modules of the project. The first module deals with the test image selection and classifying it based on the features extracted from that test image. The shape features values of that test image will be stored in another file “testfeatures.dat”. The values from this file are mapped with the values of trained neural network which are stored in the file named “net_FFBP.
mat”. Second module of second part deals with the retrieval results of the classification process. This module displays the retrieval result to the user. The first module of second part can be executed any number of times without the need of three files “features.dat” , “outtype.dat” and “net_FFBP.mat” because first module of second part involves only Segmentation and feature extraction of test image.
The second module needs those three files mentioned above because it maps the shape features of test image with the trained neural network data. So the project needs the files “features.dat”, “outtype.dat”, “net_FFBP.mat”, and “testfeatures.dat” for successful complete execution of the project.