Face recognition has a great potential to play an important role in computer vision field. However, the majority of face recognition methods are based on the low-level features, which may not yield good results. Inspired by a simple deep learning model principal component analysis network (PCANet), we propose a novel deep learning network called circular symmetrical Gabor filter (2D)2PCA neural networks [CSGF(2D)2PCANet]. Previous models used in face recognition have three major issues of data redundancy, computation time, and no rotation invariance. We introduce the CSGF to address these issues. Two-directional 2-D PCA [(2D)2PCA] is used in feature extraction stage. Binary hashing, blockwise histograms, and linear SVM are used for the output stage. The proposed CSGF (2D)2PCANet learns high-level features and provides more recognition information during the training phase, which may result in a higher recognition rate when testing the sample. We tested the proposed method on XM2VTS, ORL, AR, Extend Yale B, and LFW databases. Test results show that the CSGF (2D)2PCANet is more robust to the variation of occlusion, illumination, pose, noise, and expression, which is a promising method in face recognition.

Face Recognition Based on CSGF(2D) 2 PCANet

With the rapid development of technology, and the popularization of internet, communication is been greatly promoted. The communication is not limited only to information but also includes multimedia information like digital Images. Therefore, the security of digital images has become a very important and practical issue, and appropriate security technology is used for those digital images containing confidential or private information especially. In this paper a novel approach of Image scrambling has been proposed which includes both spatial as well as Transform domain. Experimental results prove that correlation obtained in scrambled images is much lesser then the one obtained in transformed images.

Comparative Performance of Image Scrambling in Transform Domain using Sinusoidal Transforms

The invention discloses a pupil positioning method and device and belongs to the field of electronic equipment The method comprises the steps that filtering is performed on a human eye image to be processed through the real part of a two-dimensional Gabor filter, so that a first light spot image is obtained; binaryzation and expansion treatment are performed on the first light spot image, so that a second light spot image is obtained; selective filling and Gaussian filtering treatment are performed on the human eye image to be processed according to light spots, located in a pupil, in the second light spot image, so that an evenly filled human eye image is obtained; pupil positioning is performed according to the evenly filled human eye image and a pupil positioning algorithm, so that a pupil positioning image is obtained According to the pupil positioning method and device, the light spots nearby the pupil can be detected through the real part of the two-dimensional Gabor filter, so that wrong filling is avoided; even filling of the light spots is achieved through selective filling and Gaussian filtering treatment, and then the accuracy of pupil positioning can be improved

Pupil positioning method and device

Video watermarking is relatively a new technology to ensure protection of intellectual property rights and to stop video piracy. The ownership information or watermark is normally hidden in the video sequences. In this paper, a novel video watermarking method is proposed to protect the ownership information in a robust way. This method encrypts the watermark by combining the strengths of Compressive Sensing and Arnold scrambling. The frames for watermark embedding are chosen by computing the Sum of Absolute Deviation between successive frames. The cipher watermark is then embedded into chosen frames based on SVD watermark embedding algorithm. The decryption stage performs SVD watermark extraction algorithm, Arnold inverse transform and L1 optimization for retrieval of watermark. Experimental results show that proposed method enhances the security of the ownership information embedded.

A cryptographic approach to video watermarking based on compressive sensing, arnold transform, sum of absolute deviation and SVD

The invention discloses an image scrambling evaluation method and an image scrambling evaluation device, and belongs to the field of digital image processing The image scrambling evaluation method includes respectively performing bitwise layering to the original image and at least one scrambling image to acquire a bitmap image of the original image and a bitmap image of the scrambling image, performing feature extraction to the bitmap image of the original image and the bitmap image of the scrambling image respectively according to an intersecting cortical neural network model to acquire a structural feature sequence of the original image and a structural feature sequence of the scrambling image, and calculating according to the structural feature sequence of the original image and the structural feature sequence of the scrambling image so as to acquire the scrambling degree of the scrambling image By the steps of performing the bitwise layering to the original image and the scrambling images and performing the feature extraction according to the intersecting cortical neural network model, the structural feature information of the images is extracted directly and the scrambling degree evaluation is implemented effectively

Image scrambling evaluation method and image scrambling evaluation device

Removing speckles is a key step for accurate pupil location in an iris recognition system. This paper proposed an algorithm to detect speckles in pupil area based on 2D Gabor filters first. Then the speckles were replaced selectively with the average intensity of a fixed square region. Finally, a novel evaluation index was introduced, which improved the pupil location accuracy rate of our former pupil location algorithm. In the widely used CASIA v3.0 iris database, the pupil location accuracy rate was improved from 97.44% to 99.55%. And in several other commonly used iris test databases with less or without speckles, this method maintained the former location accuracy. The experimental results show that our algorithm has satisfactory performance, robustness and versatility.

Removing Speckles Selectively from Iris Images to Improve Pupil Location Using 2D Gabor Filters

Scrambling transformation plays an important role in information hiding application, so offering an effective evaluation method for scrambling algorithms is becoming increasingly necessary. The paper firstly analyzed the Arnold transformation process to get some universal rules about the periodicity of scrambling process, then used the improved Intersecting Cortical Model Neural Network (ICMNN) designed especially to extract 1D signatures of the original image and scrambled images which could effectively reflect the image structure changing processing. Finally L1 norm was adopted to evaluate the scrambling degree and the universal rules obtained above were used to verify the results. The experimental results showed that the proposed method could analyze and evaluate the scrambling degree efficiently and had a promising application future.

Chunxian Song

Papers

Removing Speckles Selectively from Iris Images to Improve Pupil Location Using 2D Gabor Filters

Evaluation of Image Scrambling Degree with Intersecting Cortical Model Neural Network