Handwriting analysis has been addressed by researchers for decades, and many advances were achieved in understanding handwritten texts so far. However, some applications have been rarely discussed. One of these applications that has received less attention is the understanding and analyzing of handwritten circuits. Today, with the widespread use of intelligent tools in engineering and educational processes, the need for new and accurate solutions for processing such handwritings is felt more than ever. This paper presents a new method to analyze handwritten logic circuits. In this method, circuit components are first identified using a deep neural network based on YOLO. Then, the connection among these components is recognized using a new simple boundary tracking method. Then, the binary function related to the handwritten circuit is obtained. Finally, the truth table of the logic circuit is generated. We have also created a set of various handwritten logic circuits called JSU-HWLC. The results of the experiments show the proper performance of the proposed method on the collected dataset. The experiments demonstrated that the YOLO algorithm achieved better results than other deep learning methods such as faster R-CNN, Detectron2, and RetinaNet. For this reason, YOLO has been used to identify logic gates in the proposed system.

This paper addresses the efficiency of two feature extraction methods for classifying small metal objects including screws, nuts, keys, and coins: the histogram of oriented gradients (HOG) and local binary pattern (LBP). The desired features for the labeled images are first extracted and saved in the form of a feature matrix. Using three different classification methods (non-parametric K-nearest neighbors algorithm, support vector machine, and naïve Bayesian method), the images are classified into four different classes. Then, by examining the resulting confusion matrix, the performances of the HOG and LBP approaches are compared for these four classes. The effectiveness of these two methods is also compared with the “You Only Look Once” and faster region-based convolutional neural network approaches, which are based on deep learning. The collected image set in this paper includes 800 labeled training images and 180 test images. The results show that the use of the HOG is more efficient than the use of the LBP. Moreover, a combination of the HOG and LBP provides better results than either alone.