Corners detection and sensitive parts features extractions methods are gaining more interests in quality control for automated industrial sensitive products manufacturing. As 2D edge defections detections algorithms have their own limitation as they are not supportive and accurate, 3D can improve significantly the accuracy of detecting the product defects.This paper, proof the concept of the accuracy of using 3D edge defections detection in comparison with 2D. Percentages of edge defection detections have been shown to aid the decision making of accepting or rejecting the final products shaping before delivery stage. Results showed that in many cases, 3D edge defections detections aid the decision making better than 2D edge detections algorithms.

With an increasing number of security threats in recent years, the field of automatic facial recognition has seen many new developments. The introduction of many new face recognition algorithms focuses on increasing the accuracy rate of the recognition system. This paper introduces a face recognition system using Independent Component Analysis (lCA) for feature extraction and a Support Vector Neural Network (SVNN) for classification. As well as introducing a comparison between SVNN and Support Vector Machine (SVM) and Artificial Neural Network (ANN) classifiers, they are applied to prove the reliability of the proposed method. The implemented experiments use Yale databases, and the results prove that the proposed approach has a higher recognition rate than the (ICA+SVM) and (ICA+ANN) approaches for face recognition.

This study examines the possibility of remotely measuring the cardiac pulse activity of a patient, which could be an alternative technique to the classical method. This type of measurement is non-invasive. However, several limitations may deteriorate the accuracy of the results, including changes in ambient illumination, motion artifacts (MA) and other interferences that may occur through video recording. The paper in hand presents a new approach as a remedy for the aforementioned problem in cardiac pulse signals extracted from facial video recordings. Partitioning provides the basis for the presented MA reduction method; the acquired signals are partitioned into two sets for each second and every partition is shifted to the mean level and then all the partitions are recombined again into one signal, which is followed by low-pass filtering for enhancement. The proposed compared with ordinary pulse oximetry Photoplethysmographic (PPG) method. The resulted correlation coefficient was found (0.957) when calculated between the results of the proposed method and the ordinary one. Experiments were implemented using a common camera by creating a dataset from 11 subjects. The ease of implementation of this method with a simple that can be used to monitor the cardiac pulse rates in both home and the clinical environments.

Sensitive products deformation marking, and detection projects are becoming more interests in quality control management and widely used in automated industrial sensitive products production. The manual inspection by the workers for the sensitive industrial products is still taking long time and is not accurate and supportive enough for high quality control, automated deformations marking and detection by using image processing algorithms can increase highly the accuracy of detection for the defects in the products. In this paper, automated deformation detection and classification system based on some steps of image processing. Decision-making is made based on the percentage of deformations detection that aids for accept or reject the final car part product shape before deliver to users. In this work, the results were accurate to detecting the mismatching in industrial products for many cases. Automated deformations detections aid for better decision making to accepting or rejecting the products.