Zaifeng Shi

An improved anisotropic diffusion filter with semi-adaptive threshold for edge preservation

This paper presents a noise removal method based on a semi-adaptive threshold in anisotropic diffusion filter to get better detail information protection and stronger noise suppressing ability. In this model, a method of local difference value is applied to distinguish corrupted pixels and noise-free pixels, and parts of the corrupted pixels are replaced by the pixels which have been pre-denoised through a Gaussian filter. Then an anisotropic diffusion model with a semi-adaptive threshold in diffusion coefficient function is applied to get a restored image. In order to implement the semi-adaptive threshold for each diffusion, the gradient value of the corrupted pixels is introduced in the threshold, which results in more diffusion in smooth areas and less diffusion in boundary regions. Compared with the traditional anisotropic diffusion models, the experimental results show that the proposed method can improve the PSNR by 30% and SSIM by 5%. The filtered images and experimental data indicate that the proposed method performs efficiently in both edge preservation and noise removing. A new anisotropic diffusion filter based on a semi-adaptive threshold is proposed.A novel method with local difference value is applied to extract corrupted pixels.The improved method performs well in both edge preservation and noise removing.

Design of a Configurable and Extensible Tcore Processor Based on Transport Triggered Architecture

A hardware design of a configurable and extensible processor named Tcore, which is based on transport triggered architecture (TTA)¿ is presented in this paper.xa0xa0Due to its flexibility, the Tcore can be used as an application specific processor, especially as a coprocessor for different DSP applications. We have configured Tcore to an instruction level parallel processor to support the application of MP3 IMDCT in a SoC and have been fully verified on FPGA. The results show the advantage of using this configurable processor in terms of performance in computation, flexibility in application, limited effort in design and reduction on silicon area.

Adaptive defect correction and noise suppression module in the CIS image processing system

The image quality in a CIS chip is inferior to that of the CCD mainly because the random noise, which is very difficult to be suppressed in analog circuits, especially in submicron process. The popular high quality random noise suppression algorithms are too complex to be used in a color CIS chip. Besides the random noise, the defect pixels are more difficult to avoid, especially in high resolution CIS system. Therefore, a novel spacial adaptive noise suppression algorithm, which combines the defect pixel correction function, is presented. The spatially adaptive defect correction and noise suppression algorithm consists of defect pixel detector, defect value corrector, image detail estimator, and two configurable Gaussion masks to deal with different complex level image region. With a fast and effective performance, cross mask and 25-in sorting methods are applied to the defect detector and detail estimator circuits. With good PSNR and visual quality results and much less hardware cost, the proposed method can be practically used in a CIS chip or other camera systems.

Full-reference image quality assessment based on image segmentation with edge feature

Abstract Full-reference image quality assessment is widely used in many applications, such as image compression, image transmission and image mosaic. The visual masking effect has a significant impact on the perception of the human visual system, which is ignored in previous image quality assessments. Combined with the visual masking effect, a full-reference image quality assessment method by edge-feature-based image segmentation (EFS) was proposed. First, the image is segmented into three parts: contour regions, edge-extension regions and slowly-varying regions. The pixels in different regions are then described by different low-level features in the light of the visual masking effect. Finally, the low-level features in each part are pooled by two complementary aspects: visual saliency and visual masking effect. Experimental results on four large-scale benchmark databases show that the proposed method has a better prediction accuracy in all distortion types than other state-of-the-art image quality assessment indices.

Design and implementation of co-design toolset for tcore processor

Application-specific instruction processors (ASIP) tailored for the requirements are often at the center of todaypsilas embedded systems. Therefore, considerable effort has been spent on constructing tools that assist in co-designing ASIP. It is desirable that such design toolsets support an automated design flow from application source code down to synthesizable processor description and optimized machine code. In this paper, we will describe such a toolset for Tcore processor which is derived Transport Triggered Architecture (TTA). We have addressed some of the pressing shortcomings found in existing toolsets, especially the design of compiler. Finally, we present a satisfied result of an image contrast enhancement algorithm implemented using Tcore processor under many kinds of configuration through the toolset.

Dissimilar pixel counting based impulse detector for two-phase mixed noise removal

Mixed noise is a challenging noise model due to its statistical complexity. A new two-phase denoising method based on an impulse detector using dissimilar pixel counting is proposed in this paper. This method consists of two stages: detection and filtering. For the detection phase, average difference scheme is proposed to distinguish whether two neighboring pixels are similar or not, and then the number of dissimilar pixels is compared with a threshold to locate the outlier point in noisy image. An iterative framework is used for detection accuracy with the least numbers of iteration. For the filtering phase, an extended trilateral filter is used to remove the mixture of Gaussian and impulse noise, which are treated differently depending on the guidance matrix from the detection phase. Extensive experimental results demonstrate that the proposed method exhibits better noise detection capability and outperforms many existing two-phase mixed noise removal methods in both quantitative evaluation and visual quality.

A perceptual image quality index based on global and double-random window similarity

Abstract As one of indispensable steps in vision applications, the evaluation of the perceptual quality of an image attracts significant attention in recent years. This paper presents an image quality assessment algorithm using global and local similarities as two complementary aspects. It is based on two assumptions: (1) The evaluation of information in double-random local windows can capture the quality of an image accurately. (2) This method is highly consistent with human visual system in response to visual stimuli. The final quality index is obtained by combining two components, where the local quality score is calculated by comparing various features in double-random windows, and the global quality score is obtained pixel by pixel in the image. Experimental results on four large-scale benchmark databases show the effective and stable performance of the proposed method.

A real-time image stitching method based on memory space conversion

To meet the requirement of real-time processing, an image stitching method based on memory space conversion is proposed. In this method, two banks in a DDR2 SDRAM are used for storing image data alternately to realize the image stitching with image translation and rotation in hardware. The store addresses of two images are calculated based on the given stitching parameters and Bresenham algorithm. This algorithm results in smaller calculation and lower implementation cost. An arbiter scheme is proposed to meet the strict timing requirements of memory writing and reading. To ensure the timing synchronization of visual information from two image sensors, a judge module is used to determine which image should be stored first. Finally the systematic circuit simulation is conducted and the simulation results are compared with the mathematical model. Experimental results show that the proposed method not only avoids the multiplication, but also reduces the addition by 93.75%. It can improve the computation efficiency and decrease the complexity of hardware circuit on the basis of guaranteeing the image quality.

Segmented-spectrum detection mechanism for medical x-ray in CdTe

This paper presents a segmented X-ray spectrum detection method based on a layered X-ray detector in Cadmium Telluride (CdTe) substrate. We describe the three-dimensional structure of proposed detector pixel and investigate the matched spectrum-resolving method. Polychromatic X-ray beam enter the CdTe substrate edge on and will be absorbed completely in different thickness varying with photon energy. Discrete potential wells are formed under external controlling voltage to collect the photo-electrons generated in different layers, and segmented X-ray spectrum can be deduced from the quantity of photo-electrons. In this work, we verify the feasibility of the segmented-spectrum detection mechanism by simulating the absorption of monochromatic X-ray in a CdTe substrate. Experiments in simulation show that the number of photo-electrons grow exponentially with the increase of incident thickness, and photons with different energy will be absorbed in various thickness. The charges generated in different layers are collected into adjacent potential wells, and collection efficiency is estimated to be about 87% for different incident intensity under the 40000V/cm electric field. Errors caused by charge sharing between neighboring layers are also analyzed, and it can be considered negligible by setting appropriate size of electrodes.

Adaptive Partition-Cluster-Based Median Filter for Random-Valued Impulse Noise Removal

As the most popular nonlinear denoise technique, the median filter has attracted significant attention in recent years. In this paper, a novel adaptive median filter is presented to remove random-valued impulse noise in images, named Adaptive Partition-Cluster-Based Median (APCM) Filter. Based on the partition cluster idea, the noise detector classifies pixels into different groups and identifies the noisy pixels in different regions adaptively without iterations. According to the results of noise detection, an improved adaptive decision-based filter is presented to restore the pixels which are corrupted by random-valued impulse noise. The proposed filter technique is open to any impulse noise. Extensive simulation results demonstrate that the proposed method substantially outperforms other state-of-the-arts impulse noise filter techniques both visually and in terms of objective quality measures. Furthermore, the proposed method is much friendly to the hardware parallel implementation of image processing ...