Skin cancer is one of the most common types of cancer, whose number of cases is constantly increasing. The most used method to detect skin cancer is the biopsy. It is relevant to reduce the number of biopsies, since it is an invasive and expensive procedure, and it has limited availability in some locations. Among the most successful approaches that aim to improve skin cancer detection are the algorithms that process active infrared thermography. Here, a skin cancer detection scheme is proposed, which extracts key features from active thermography videos, and uses them in the following five classifiers: K-Nearest Neighbors, Decision tree, Random forest, Support vector machine (SVM) and extrem e Gradient Boosting (XGBoost). Under a minimization error design criteria, the best result was performed by a SVM classifier, reaching 84.14% of accuracy and 78.92% of precision. Modifying the classifiers to ensure that all the malignant cases are detected, the best performance was also achieved by the SVM classifier, with 72.85% of accuracy and 63.95% of presicion. The proposed scheme is 15% less accurate than the best detection algorithm. However, it is easier to implement and deploy and provides a framework with key preprocessing aspects to address this detection problem using active thermography. As future work, a further exploration of features will be carried out, with the aim of improving the performance of the classifier.