Advanced Materials Research Vols. 760-762

Abstract: This paper studies the pre-processing methods on Local Binary Pattern (LBP) for face recognition. Three methods are well investigated, such as Gaussian smoothing, histogram equalization and Sobel operator. The extensive experiments on FERET database show that the best recognition rate increase is achieved by combination of histogram equalization, Gaussian smoothing, and Sobel with LBP. The conclusion achieved in the paper is useful for real face recognition system.

Abstract: For the channel source of large capacity image data, the error correcting capability and coding efficiency of channel encoding is very important, in order to solve the real-time and parallel of encoding, a kind of code encoding method realized easily by FPGA is proposed. First, the coding principle of the convolutional code is introduced in detail; Secondly, the representation method of the convolutional encoder is elaborated, including the state transition diagram and the grid diagram of the convolutional code; Then, the convolutional encoding algorithm based on FPGA will be converting to achieve rapid, parallel processing method; Finally, (2,1,7) convolutional code algorithm is simulated and tested. The experimental results show that: the convolutional encoding module is capable of handling the input data stream of up to 160Mbps, processing speed, to meet the parallel and real-time of convolutional encoding for the channel source of large capacity image data, to improve the efficiency of convolutional encoding.

Abstract: License plate location is the first and crucial stage in license plate recognition. Although many license plate location methods achieve higher success rate, they are always time-consuming. However, robust real-time characteristic is important to application of license plate recognition. In this paper, a texture-based method of license plate location is proposed to enhance real-time characteristic while keeping higher success rate. It consists of three parts: preprocessing, region of interested (ROI) location and ROI filtration. Experimental has demonstrated that the proposed method achieve average execution time of 49ms and success rate of 96.3% simultaneously.

Abstract: Finger-vein recognition technology is a new contactless biometric identification technology, which has a good prospect for development. This paper presents an algorithm of recognition and matching of ROI region of finger-vein based on NMI feature. The algorithm references ROI conception in the image preprocessing, and uses NMI feature as the image feature value in the feature recognition. The paper sets up five experiments to study and analyze the performance of the algorithm, the results has showed that this algorithm assumes high efficiency, good stability and reliability with simple implementation procedure.

Abstract: Low-resolution Chinese character recognition of license plate is always a difficult problem. For solving it, we must think about the distinctiveness of character feature and the counting speed of method simultaneously. In this paper, we proposed a simple and effective feature extraction algorithm. First, extract the statistical feature of Chinese character based on decomposing stroke with wavelet transform. Second, apply Elastic Mesh Algorithm into extracting wavelet coefficient of decomposing stroke to get the structure information of Chinese character. The experimental results show the method is robust against low quality Chinese characters, such as skew, fuzzy, glue, distorted character, and easy to be used in actual projects with simple advantage.

Abstract: Multi-thresholding is an important step for automatic image analysis. In this paper, a multi-thresholding method based on symmetric Tsallis-cross entropy and uniform searching particle swarm optimization (UPSO) is proposed. The criterion function using symmetric Tsallis-cross entropy can make the grayscale within the background cluster and the object cluster uniform. Since the exhaustive multi-thresholding algorithm would be too time-consuming, UPSO algorithm is adopted to find the optimal thresholds quickly and accurately. A large number of experimental results show that, compared with related multi-thresholding methods based on Shannon entropy and Tsallis entropy, the proposed method is effective and rapid. It can obtain more accurate boundary shape and clearer details of object.

Abstract: To further improve the accuracy of SAR image segmentation in the marine spill oil detection, a segmentation method of marine spill oil images based on Gabor, Krawtchouk moments and KFCM is proposed in this paper. Firstly, the marine spill oil image is decomposed by Gabor transform to obtain the texture features of image. Then, the Krawtchouk moments are applied to extract the shape features of image. Finally, the image segmentation is achieved based on KFCM. A large number of experimental results show that, compared with the related segmentation methods such as Tsallis entropy threshold method,CV model method and the method based on Gabor, Krawtchouk moments and FCM, the proposed method can achieve better result.

Abstract: Aiming at welding defect image with complex background and low contrast, a segmentation method of welding defect image based on exponential cross entropy and improved pulse coupled neural network (PCNN) is proposed. Firstly, the area of weld is extracted by gray projection algorithm. Then, link weighted matrix and dynamic threshold function of PCNN are improved. Finally, the exponential cross entropy is calculated as criterion to determine the number of iteration for improved PCNN and get the optimal segmented image. The experimental results are given. Compared with the threshold segmentation method based on exponential cross entropy, the segmentation method based on PCNN and Shannon entropy, the proposed method can achieve better segmented results.

Abstract: Segmentation of defect images is an important step in the automatic fabric defect detection. In order to extract fabric defects effectively, a segmentation method of fabric defect images based on pulse coupled neural network (PCNN) model and symmetric Tsallis cross entropy is proposed. The image is segmented by PCNN according to the gray strength difference between fabric defect area and non-defect area. To guarantee that the grayscale inside the object and background is uniform after segmentation, symmetric Tsallis cross entropy is used as the image segmentation criterion to select the optimal threshold and iteration number. A large number of experimental results show that, compared with the related segmentation methods such as Otsu method, PCNN method, the method based on PCNN and cross entropy, the segmentation effect of the proposed method is the best. The texture of non-defect area is removed more completely, and the defect area is segmented more accurately.

Abstract: In order to improve the accuracy and computational efficiency of change detection of multi-temporal remote sensing images, a change detection algorithm based on nonsubsampled contourlet transform (NSCT) and independent component analysis (ICA) is proposed. The flexibility of NSCT in image decomposition and the effectiveness of ICA in image separation are used comprehensively. Firstly, multi-scale decomposition of remote sensing images is performed by NSCT. Then the decomposed low-frequency components and high-frequency components form into partitioned vectors. ICA is carried out for the partitioned vectors and separates mutual independent components. Next the separated components are transformed into image components which include the change image. Finally, change detection result is achieved by threshold segmentation and filtering of the change image. The experimental results show that, compared with the algorithm based on ICA, the algorithm based on wavelet transform and ICA, the proposed algorithm separates change information more effectively and reduces computational complexity. The obtained change image has higher accuracy and stronger robustness to the background.