This thesis implemented a field programmable gate array (FPGA)-based face detector using a neural network (NN), as well as a bit-width reduced floating-point unit (FPU). An NN was used to easily separate face data and non-face data in the face detector. The NN performs time consuming repetitive calculation. This time consuming problem was solved by a Field Programmable Gate Array (FPGA) device and a bit-width reduced FPU in this thesis. A floating-point bit-width reduction provided a significant saving of hardware resources, such as area and power.<p>The analytical error model, using the maximum relative representation error (MRRE) and the average relative representation error (ARRE), was developed to obtain the maximum and average output errors for the bit-width reduced FPUs. After the development of the analytical error model, the bit-width reduced FPUs and an NN were designed using MATLAB and VHDL. Finally, the analytical (MATLAB) results, along with the experimental (VHDL) results, were compared. The analytical results and the experimental results showed conformity of shape. It was also found that while maintaining 94.1% detection accuracy, a reduction in bit-width from 32 bits to 16 bits reduced the size of memory and arithmetic units by 50%, and the total power consumption by 14.7%.

This paper describes my participation in a MITRE Corporation sponsored computer science clinic project at Harvey Mudd College as my senior project. The goal of the project was to implement a landmark-based pose correction system as a component in a larger, existing face recognition system. The main contribution I made to the project was the implementation of the Active Shape Models (ASM) algorithm; the inner workings of ASM are explained as well as how the pose correction system makes use of it. Included is the most recent draft (as of this writing) of the final report that my teammates and I produced highlighting the year's accomplishments. Even though there are few quantitative results to show because the clinic program is ongoing, our qualitative results are quite promising.

No description available.

The accurate integration of audio-visual emotion cues is critical for social interactions and requires efficient processing of facial cues. Gaze behavior of typically developing young adults was measured via eye-tracking during the perception of dynamic audio-visual emotion (DAVE) stimuli. Participants were able to identify basic emotions (angry, fearful, happy, neutral) and determine the congruence of facial expression and prosody. Perception of incongruent videos resulted in increased reaction times and emotion identification consistent with the facial expression. Participants consistently demonstrated a featural processing approach across all tasks, with a significant preference for the eyes. Evidence of hemispheric lateralization was indicated by preferential fixation to the left (happy, angry) or right eye (fearful). Fixation patterns differed according to the facially expressed emotion, with the pattern that emerged during fearful movies supporting the significance of automatic threat processing. Finally, fixation pattern during the perception of incongruent movies varied according to task instructions.

Face detection is a classical application of object detection. There are many practical applications in which face detection is the first step; face recognition, video surveillance, image database management, video coding. This report presents the results of an implementation of the AdaBoost algorithm to train a Strong Classifier to be used for face detection. The AdaBoost algorithm is fast and shows a low false detection rate, two characteristics which are important for face detection algorithms. The application is an implementation of the AdaBoost algorithm with several command-line executables that support testing of the algorithm. The training and detection algorithms are separated from the rest of the application by a well defined interface to allow reuse as a software library. The source code is documented using the JavaDoc-standard, and CppDoc is then used to produce detailed information on classes and relationships in html format. The implemented algorithm is found to produce relatively high detection rate and low false alarm rate, considering the badly suited training data used.

This thesis deals with planar drawings of planar graphs such that each interior face has a prescribed area. Our work is divided into two main sections. The rst one deals with straight-line drawings and the second one with orthogonal drawings. For straight-line drawings, it was known that such drawings exist for all planar graphs with maximum degree 3. We show here that such drawings exist for all planar partial 3-trees, i.e., subgraphs of a triangulated planar graph obtained by repeatedly inserting a vertex in one triangle and connecting it to all vertices of the triangle. Moreover, vertices have rational coordinates if the face areas are rational, and we can bound the resolution. For orthogonal drawings, we give an algorithm to draw triconnected planar graphs with maximum degree 3. This algorithm produces a drawing with at most 8 bends per face and 4 bends per edge, which improves the previous known result of 34 bends per face. Both vertices and bends have rational coordinates if the face areas are rational.

Class III malocclusion in orthodontic patients typically results from midfacial retrognathia. However, the etiology of the midfacial retrognathia remains unclear. The cranial base is considered to play an important role in the emergence of midfacial morphology due to its location within the craniofacial region as well as its dramatic growth activity during the later prenatal and early postnatal periods. Previously, the nasal septum was considered a key cranial base component which functioned to pull the maxilla anteriorly during growth. Although this nasal septal theory of midfacial advancement is generally accepted, certain midfacial abnormalities occur in the presence of normal nasal septal morphology and growth indicating that additional craniofacial regions must contribute to the control of midfacial prognathism. The purpose of this study was to describe a mouse mutant which displays midfacial retrognathia and to delineate regions of cranial base malgrowth. Further, cellular growth mechanisms responsible for causing the abnormal cranial base growth trajectories were identified. Adult 3H1 Brachyrrhine (Br) male mice, displaying midfacial retrognathia with a characteristic Class III malocclusion, were bred to normal C3H females. Litters were examined to determine whether Br offspring could be distinguished from one another between Theiler stages 23 (E15) and 27 (E19) using qualitative and quantitative methods. Results showed that two distinct groups of offspring were derived: one with midfacial retrognathia and the other without. The cranial base of Br mutants displayed a malformed sphenoidal region while the nasal septum appeared much less affected as revealed by finite element morphometric analysis. In vivo autoradiographic analysis demonstrated the existence of temporal growth sites (TGS). TGS in the sphenoidal regions were deficient in the prenatal Br mouse. Using immunohistochemistry, insulin growth factor (IGF-I) and epidermal growth factor (EGF) as well as their receptors (IGF-IR and EGF-R) were found to be expressed in cranial base chondrocytes. In order to determine whether chondrocytic responses to these growth factors were deficient in the mutant, cell cultures were established, treated with IGF-I or EGF, and cellular proliferation and differentiation were measured using (³H) -thymidine and (³⁵S) -sulfate incorporation. Results from this analysis showed that chondrocytes from the Br posterior cranial base were less responsive to EGF compared to cells from normal posterior cranial bases. Data from this study suggest that the Br mouse displays midfacial retrognathia in a heritable fashion. Deficient growth of the Br sphenoid, particularly in the presphenoidal and sphenoethmoid region, is crucial for the establishment of midfacial retrognathia. In the Br mouse, the sphenoidal deficiency is associated with depressed proliferation in TGS. IGF-I and EGF, as well as their receptors, are expressed within the murine cranial base and the depressed level of cellular proliferation in the Br sphenoidal region results, at least in part, from a diminished response to EGF. Based on this study, I propose a "sphenoidal mechanism of midfacial advancement" whereby the sphenoid actively propels the midface forward in order to achieve proper maxillary prognathism during later prenatal and early postnatal period.

This thesis implemented a field programmable gate array (FPGA)-based face detector using a neural network (NN), as well as a bit-width reduced floating-point unit (FPU). An NN was used to easily separate face data and non-face data in the face detector. The NN performs time consuming repetitive calculation. This time consuming problem was solved by a Field Programmable Gate Array (FPGA) device and a bit-width reduced FPU in this thesis. A floating-point bit-width reduction provided a significant saving of hardware resources, such as area and power.<p>The analytical error model, using the maximum relative representation error (MRRE) and the average relative representation error (ARRE), was developed to obtain the maximum and average output errors for the bit-width reduced FPUs. After the development of the analytical error model, the bit-width reduced FPUs and an NN were designed using MATLAB and VHDL. Finally, the analytical (MATLAB) results, along with the experimental (VHDL) results, were compared. The analytical results and the experimental results showed conformity of shape. It was also found that while maintaining 94.1% detection accuracy, a reduction in bit-width from 32 bits to 16 bits reduced the size of memory and arithmetic units by 50%, and the total power consumption by 14.7%.

Based on the theory of Spiral of Silence from Noelle-Neumann, the main purpose of this research is to figure out internet users how to express their opinion after evaluating opinion climate and whether the characters of the Internet make their opinion- expression arbitrarily. Moreover, this research focuses on Chinese ¡¥face¡¦ culture to examine the cultural context how to influence opinion- expression of internet users. This research chooses ¡¥PeoPo¡¦ citizen journalism platform as research field, and adopts in-depth interview to analyze the opinion- expression of 9 PeoPo citizen reporters. As for the research topic, it¡¦s concerned with the issue of village migration of Taiwanese aborigine after the disaster caused by Typhoon Morakot in 2009. Therefore, this research attempts to figure out the PeoPo citizen reporters¡¦ inner process of opinion- expression and the influence of Chinese face culture. This research finds that the opinion-evaluation of internet users mainly rely on the Internet, and the fear of isolation that the spiral of silence theory emphasizes don¡¦t predict their willingness to speak up alone, but Chinese face culture does. Furthermore, the characters of the Internet don¡¦t prevent the spiral of silence effect. Although PeoPo not only accepts diverse viewpoints widely, but also inspires more citizens to voice local issues, the results show that actually PeoPo citizen reporters express their opinion with great pressure mostly lowering their willingness to speak up such as the foundation idea of PeoPo, reference group, threat of opposition, objective principle of journalism, and Chinese face culture. In sum, getting one¡¦s own way is discouraged on PeoPo; in other words, the characters of the Internet still unable to relieve PeoPo citizen reporters of speaking up.

The thesis illustrates the development of DSP-based¡§Real-Time Human Face Recognition System¡¨.The principal system consists of three major subsystems.There are Image Acquisition System¡AImage Preprocessing System and human face characteristic extraction . For experiment,we adopted colored face image with complex background and simulate on PC.We found the characteristic points and characteristic vectors from the face image which is searched from Gene algorithm.Then,we use the recognition system to recognize the face image.Finally we implant it to DSP. Shown by the experimental result,this system has good recognition and efficiency.