Applied Mechanics and Materials Vols. 411-414

Abstract: A method for precise face detection and segmentation is presented combining with the particle swarm optimization. In order to extract face feature precisely, the method uses two key technologies: skin color segmentation and particle swarm search. The experimental results show that this method can eliminate the interference factor, and improve the accuracy of expression recognition, especially well for some single expression recognition.

Abstract: In a passive millimeter wave (PMMW) imaging system, the resolution of the acquired image is limited by the antenna size. The Richardson—Lucy (RL) algorithm is a simple and nonlinear method, which can improve the resolution of the image. However, when the noise can not be neglected, it is difficult for RL algorithm to get good restoration of the corrupted image. To the best of our knowledge, the block-matching with 3D transform domain collaborative filtering (BM3D) algorithm achieves very good performance in image de-noising. In order to improve the resolution of passive millimeter wave images, a RL imaging algorithm for passive millimeter wave based on BM3D is proposed in this paper. The modified algorithm effectively reduces the influence of noise on RL algorithm by using de-noise algorithm based on BM3D. Experimental results demonstrate that the proposed algorithm improves the performance of RL algorithm. Furthermore, the algorithm can be easily implemented for passive millimeter wave imaging．

Abstract: In terms of the phenomenon that red cells and white cells incline to touch each other in urinary sediment images, an improved touching-cell splitting algorithm is proposed in the paper that is based on the bottleneck detection The proposed algorithm is featured by positioning the bottleneck points by the rule of bottleneck, also featured by determining to split a region or not by the criteria of segmentation conditions. Once a splitting is done, the algorithm turn to apply the bottleneck detection again to those regions that are already been segmented. Simulation and experiment show that the performance of the improved algorithm is accurate and stable in splitting ranging from two touched cells to multiple touched circumstances; the enhanced robustness and universality can be enough to prove the practicability of the algorithm a certain level.

Abstract: Image deblurring with noise is a typical ill-posed problem needs regularization. Various regularization models were proposed during several decades study, such as Tikhonov and TV. A new regularization model based non-local similarity constrains is proposed in this paper, which used l² non-local norms and could be easily solved by fast non-local image denoising algorithm. By combining with Bregmanrized operator splitting (BOS) algorithm, a fast and efficient iterative three step image deblurring scheme is given. Experimental results show that proposed regularization model obtained better results on ten common test images than other similar regularization model including newly proposed NLTV regularization, both in deblurring performance (PSNR and MSSIM) and processing speed.

Abstract: In allusion to randomness and fuzziness of digital image semantic, we propose a new semantic representation of digital image based on cloud model and construct a semantic vector space. In this space, semantic classifications of digital images are completed by calculating the semantic class certainty degree (SCCD). In addition, we propose cloud support vector machine based on image semantics (CSVM-IS) model. Experimental results show that CSVM-IS can accomplish target classification and has good classification accuracy.

Abstract: In this paper, a novel crack detection method is proposed based on the digital image of ancient building painting. The paper first does painting image preprocessing, such as image edge detection, image binary of adaptive threshold and removal of isolated points, and then establishes discrimination model to distinguish between regions and lines, finally uses curve fitting way to sort out the cracks in building painting. The experimental results are satisfactory.

Abstract: Maximum Scatter Difference (MSD) aims to preserve discriminant information of sample space, but it fails to find the essential structure of the samples with nonlinear distribution. To overcome this problem, an efficient feature extraction method named as Locality Preserving Maximum Scatter Difference (LPMSD) projection is proposed in this paper. The new algorithm is developed based on locality preserved embedding and MSD criterion. Thus, the proposed LPMSD not only preserves discriminant information of sample space but also captures the intrinsic submanifold of sample space. Experimental results on ORL, Yale and CAS-PEAL face database indicate that the LPMSD method outperforms the MSD, MMSD and LDA methods under various experimental conditions.

Abstract: In this paper, a JPEG steganalysis algorithm based on locality preserving projection (LPP) dimensionality reduction method is proposed for detecting the unseen stego algorithms. The co-occurrence features are extracted from DCT-DWT domain and dimension is reduced by using the LPP method. For improving the generalization capability of the algorithm, SVDD is used as the classifier. Experimental results reveal the fact that our scheme has better generalization capability and is more effective than others.

Abstract: In this paper, a new adaptive images fusion algorithm is presented for CT and MRI based on DT-CWT. For fusion, all the source images are decomposed into low and high frequency sub-bands, and then the fusion of low frequency is done by means of Principal Component Analysis (PCA) while for high frequency regional energy algorithm is used. Experiment are carried out on a number of CT and MRT images, results show that the DT-CWT method is better than that of DWT method in terms of quality measures PSNR, NCC and image visual quality.

Abstract: In this paper, we describe a novel encryption algorithm, which converts a greyscale image into a colored JPEG image. Firstly, it creates MCU (Minimum Coding Unit) of the colored JPEG image from the DU (Data Unit) of the greyscale image by the 8x8 construction matrix randomly. Secondly, it shuffles all the DUs with quantized DCT (Discrete Cosine Transform) coefficients according to a random ergodic matrix. Lastly, it rearranges the DUs as the format of the colored JPEG image and proceeds with the normal compression and encoding. The results show that the encryption speed of the algorithm is fast enough for real-time transmission and the encrypted image has almost the same size as original image after direct compression.