Activity 18: Pattern Recognition

In this activity we are to identify to what class an object belongs to. We do this by extracting relevant features about different samples of the object and comparing its features to a training set. The criteria for membership in a certain class is when its features has a minimum distance to a certain class' representative features obtained by getting the mean of each features of the samples from each class. The classes I used are piatos chips, pillow chocolate filled snack, kwek kwek and squidballs. The images of the objects are shown below:



I chose four features that can be extracted from the objects; normalized area obtained via pixel counting, average value of the red component, ratio of the average value of the green component to the red component, and the ratio of the average value of the blue component to the red component. Since we have only eight samples I used the first four as a training set and the tested the class membership of the training set. The mean feature vectors are shown below:



Class membership is obtained by taking the minimum distance between a test's feature to the mean of each feature, following the formula given by Dr. Soriano's pdf file:



Using the features and the above formula, I obtained 15 correct classifications out of 16 samples translating to a 93.75% correct classification.

For the extraction of the parameters of the samples, I performed parametric color image segmentation we learned from activity 16. Afterwards, I inverted the segmented image, performed the morphological closing operation and multiplied it to the original colored image to obtain a colored segmented image. I used excel to perform the minimum distance classification, it was tedious but I understand it better if I do it this way :0. The scilab code I used for the extraction of the parameters is shown below:

For this activity I grade myself 10/10 since I was able to get a relatively high correct classification percentage.

Ed David helped me a lot in this activity. Special thanks to Cole, JM, Billy, Mer, Benj for buying the samples for this activity.