Offline optical-character-recognition is considered part of the significant pattern-recognition applications. Handwriting recognition though is an active and challenging area of research in this field. Nowadays, many algorithms are introduced with the aim of achieving the highest accuracy. To achieve a better recognition result, the input character images must have good quality. That is why the preprocessing step becomes essential for any image identification task. As for the classification phase, some researchers focused their research on the utilization of CNN (Convolutional Neural Networks). CNN algorithm is equipped with various learning architectures, enabling researchers to choose the most effective architecture for classification. However, this study suggests that the preprocessing mechanism is also another important factor that should be considered in order to increase the accuracy of classification. This study utilized seven stages for preprocessing mechanisms applied to the selected dataset and the measurement of the accuracy as the result of the utilization of the CNN for the classification was acquired. The study exposed that the utilization of using preprocessing steps indicated the most heightened accuracy with Barbarized-KDS as an input achieving training, testing, and validation accuracy of 0.992, 0.97, and 0.972 respectively after only 35 iterations. Furthermore, another significant finding is that using different steps in the processing of input images to trian the recognition model affects the recognition-rate within the same classifiers. Besides, the outcome reveals that each technique applied with the specific classifier may require a certain preprocess to obtain its optimal accuracy recognition rate.