Publication

Abstract : Food is a fundamental component for the physical and mental well-being of a human being. Globally, various spices are employed to make food palatable. Of the common spices, such as turmeric powder, chilli powder, masala powder, and the like, chilli powder occupies a major share in Indian cuisine. The quality of food that humans consume is of utmost significance. Several food products and other substances are added as adulterants to chilli powder like brick powder, red beetroot, tomato peel and others to give better appearance, to get better profits and several other reasons. As brick powder forms a major share of the adulterant and is harmful to the human body, this work focuses on determining the various proportions of chilli powder in specific quantity of chilli powder using image processing. A dataset prepared for the work with pure chilli powder. Various color spaces like RGB, YCbCr, HSV and LAB are employed on pre-processing by average filter to get specific texture properties. These features are selected using Spearman's correlation and processed by four ML models namely Random Forest, DecisionTree, SVM, andKNN.By employing the performance indices accuracy, F1 score, recall and precision, Random Forest is observed to outperform compared with other classifiers. The same was compared with the conventional Ultrasound extraction also. The work gave some openings to adulteration analysis using deep learning, which can be attempted with more samples. A final Mobile App is being built to test the adulteration of the chilli powder by the women at home itself.