In the literature, the recognition of human facial emotions has been greatly improved. However, the big challenge is to train efficient and fast algorithms due to the large volume of input data. This article presents a novel method to investigate seven human classic emotions: angry, disgusted, happy, neutral, sad, surprised, and afraid. Instead of using information vectors of all pixels extracted from the image, the present study works with 7 Euclidean distances considering keypoints of the face: corners of the mouth, eyes, and tip of the nose, in addition to the mean and variance of the face region, 9 input attributes, in total. In addition, we verified through neurogenetic algorithms whether it is possible to use one-dimensional data as information carriers for the classification of facial emotions. Two classification algorithms were performed separately: Support Vector Machine and Multi-Layer Perceptron. Then, the genetic algorithm was applied to select the most suitable input attributes for each architecture. With this, we analyzed which variables were relevant for the classification. The database of facial emotions used were The Japanese Female Facial Expression (JAFFE) and Extended Cohn-Kanade Dataset (CK+) containing 188 and 902 images, respectively. The best result was found for the MLP architecture with 3 input attributes selected. The global average accuracy found was 80.9%, being 100% and 95% for the prediction of happy and neutral emotions, respectively.