Pneumonia, a major cause of child mortality below five years, requires a quick and accurate diagnosis from chest X-rays. However, traditional methods can be time consuming, inconsistent, and subjective. Deep learning approaches have identified intricate details and patterns in the images and extracted essential features. These methods are pulled back by complexity and high computational overhead, making it difficult to use in constrained environments. This work uses Median Filter and Histogram Equalization to improve quality and eliminate the noise of the chest X-rays. Features were extracted using the SIFT, LBP, GLCM, and HOG techniques. Simple neural network and machine learning models were trained for binary classification. With an accuracy of 98.92% on the SIFT dataset, SVM surpassed the previous research by 1.02%. While the SVM performed best for the SIFT dataset, the HOG achieved the overall best performance for all models.