Applied Mechanics and Materials Vols. 433-435

Abstract: 3D video can bring better viewing experience than traditional one, and in recent years has been developing rapidly. But the caption viewing has become one of the elements that impact on the viewing effect as well as the causes of visual fatigue. This paper introduces the overlay process of 3D subtitles, and carefully analyzed and summarized 3 D subtitles technology research from both domestic and overseas and finally summarized its proposed future research directions.

Abstract: Two kinds of imperfections, namely the detection errors and the asynchrony between phonological attributes and phone boundaries, can cause a substantial decline in recognition accuracy of a detection-based automatic speech recognition system. To solve these problems, an adjustment method between phonological attributes and phone boundaries is proposed in this paper. At first the prior knowledge of corpus and the detection results are combined, then the asynchronies in the phone boundary area are compensated and the detection errors are corrected; additionally, by selectively deleting some frames with errors, the precision of the phone models are improved. After adoption of this adjustment method, 1.4% of phoneme recognition rate can be improved in the TIMIT phone classification experiments based on Conditional Random Fields.

Abstract: Abastruct. Compressive sensing is a novel signal sampling theory under the condition that the signalis sparse or compressible.In this case,the small amount of signal values can be reconstructed when signal is sparse or compressible.This paper has reviewed the idea of OMP,GBP and SP,given algorithms and analyzed the experiment results,suggested some improvements.

Abstract: An efficient multi-hypothesis Temporal Error Concealment (TEC) method is proposed in this paper. Two compensated MarcoBlocks (MB) are obtained by two basic TEC approaches. Process of noise Motion Vector (MV) removal is introduced to get compensated MB more accurately. The lost MB is then concealed by weighted combination of the two MBs in which the weight is adjusted adaptively by calculating standard deviation of candidate MV set. Experimental results demonstrate that the proposed method can overcome shortcoming of fixed weight and provide better performance.

Abstract: In order to get better image Semantics recognition, in this paper, a scene Semantics recognition system based on fuzzy reasoning is presented. The system contains three parts: image preprocessing, target recognition, and fuzzy reasoning machine. Compared with other methods, the outputs of pattern classifiers are fuzzed, the fuzzy relationships between targets are extracted, and fuzzy inference is performed using fuzzy automata. The experiment indicates that this method could overcome the problems of false positive and false negative of pattern classifiers, and perform relatively more accurate image semantics recognition than other existing methods.

Abstract: A sparse auto-encoder model was trained to extract the code of different facial expression, which comprises four encoder layers and three decode layers, the representation locating in the fourth layer (code layer) is the features expected. With large amounts of patches randomly selected from training faces, the model was trained firstly via backpropagation which minimizes an unsupervised sparse reconstruction error, and then a softmax classifier was learned for supervised classification. The input vector for the classification is the feature of facial image induced by the learned sparse auto-encoder and two key operations (convolving and pooling). Using a small number of hidden units per layer and a relatively small number of training set, the proposed model achieves excellent performance in the experiments.

Abstract: In this paper, based on the value measure of the Intermediary truth value, the Intermediary filtering algorithm is applied to the thenar palmprint image preprocessing. By using the Objective indicators of peak signal to noise ratio (PSNR), it can be seen that compared with the classical de-noising algorithm, the intermediary filtering algorithm is more effective and practicable.

Abstract: A new method was proposed to solve match and segmentation problem of ultrasonography of the fetus brain for screen the fetal brain malformations. First, obtaining the gray value of the brain of skull, lateral ventricle (LV), and cerebella hemisphere (CH) based on the image process. Then, index of the each parts gray value scope of the health fetus brain important regions were calculated by using the edge detection based random ellipse detection (RED), using the level set method for the segmentations in tested tissues. Mean values of all datasets were calculated and a standard model were established. This standard model can be used to match the gray level of the undiagnosed groups in order to screen the fetal brain malformations. The propose method gets encouraging result of the application in 3 fetuses with hydrocephalus.

Abstract: A FCM-based segmentation algorithm is proposed in this paper to improve the accuracy and efficiency of liver parenchyma segmentation. The proposed segmentation method consists of four steps as follows:First,we characterized the gray distribution of the unfiltered image. Second, combined with the Otsu algorithm and associated with a cropped liver image, we defined a gray interval as the livers intersity range. Third, The fuzzy c-means clustering algorithm was applied to define the confidence interval of traditional confidence connectivity method. Finally, we employed the improved confidence connected algorithm to extract the liver parenchyma from a large cross-section liver image. Experimental results show that the proposed segmentation method is feasible even for diseased liver images.

Abstract: In this paper, we present a new detection algorithm for tower crane. First the objects are segmented by an improved J-SEG algorithm, and then the geometric characteristics and luminance information are used to identify the tower crane objects. Experimental results indicate our algorithm can deal with most of the tower crane images, and it is suitable for the track object initialization in the video surveillance.