Intravascular optical coherence tomography (IVOCT) is rapidly becoming the method of choice for the in vivo investigation of coronary artery disease. While IVOCT visualizes atherosclerotic plaques with a resolution <20µm, image analysis in terms of tissue composition is currently performed by a time-consuming manual procedure based on the qualitative interpretation of image features. We illustrate an algorithm for the automated and systematic characterization of IVOCT atherosclerotic tissue. The proposed method consists in a supervised classification of image pixels according to textural features combined with the estimated value of the optical attenuation coefficient. IVOCT images of 64 plaques, from 49 in vivo IVOCT data sets, constituted the algorithm’s training and testing data sets. Validation was obtained by comparing automated analysis results to the manual assessment of atherosclerotic plaques. An overall pixel-wise accuracy of 81.5% with a classification feasibility of 76.5% and per-class accuracy of 89.5%, 72.1% and 79.5% for fibrotic, calcified and lipid-rich tissue respectively, was found. Moreover, measured optical properties were in agreement with previous results reported in literature. As such, an algorithm for automated tissue characterization was developed and validated using in vivo human data, suggesting that it can be applied to clinical IVOCT data. This might be an important step towards the integration of IVOCT in cardiovascular research and routine clinical practice.
(100.0100) Image processing; (100.2960) Image analysis; (100.4995) Pattern recognition, metrics; (170.0170) Medical optics and biotechnology; (170.6935) Tissue characterization
Follicular Lymphoma (FL) is one of the most common non-Hodgkin Lymphoma in the United States. Diagnosis and grading of FL is based on the review of histopathological tissue sections under a microscope and is influenced by human factors such as fatigue and reader bias. Computer-aided image analysis tools can help improve the accuracy of diagnosis and grading and act as another tool at the pathologist’s disposal. Our group has been developing algorithms for identifying follicles in immunohistochemical images. These algorithms have been tested and validated on small images extracted from whole slide images. However, the use of these algorithms for analyzing the entire whole slide image requires significant changes to the processing methodology since the images are relatively large (on the order of 100k × 100k pixels). In this paper we discuss the challenges involved in analyzing whole slide images and propose potential computational methodologies for addressing these challenges. We discuss the use of parallel computing tools on commodity clusters and compare performance of the serial and parallel implementations of our approach.
Follicular Lymphoma; Immunohistochemical; CD10; K-means; Parallel Computing
High-throughput evaluation of tissue biomarkers in oncology has been greatly accelerated by the widespread use of tissue microarrays (TMAs) and immunohistochemistry. Although TMAs have the potential to facilitate protein expression profiling on a scale to rival experiments of tumour transcriptomes, the bottleneck and imprecision of manually scoring TMAs has impeded progress.
We report image analysis algorithms adapted from astronomy for the precise automated analysis of IHC in all subcellular compartments. The power of this technique is demonstrated using over 2000 breast tumours and comparing quantitative automated scores against manual assessment by pathologists.
All continuous automated scores showed good correlation with their corresponding ordinal manual scores. For oestrogen receptor (ER), the correlation was 0.82, P<0.0001, for BCL2 0.72, P<0.0001 and for HER2 0.62, P<0.0001. Automated scores showed excellent concordance with manual scores for the unsupervised assignment of cases to ‘positive' or ‘negative' categories with agreement rates of up to 96%.
The adaptation of astronomical algorithms coupled with their application to large annotated study cohorts, constitutes a powerful tool for the realisation of the enormous potential of digital pathology.
image analysis; immunohistochemistry; breast cancer; systems pathology; digital pathology
Image labeling is an essential step for quantitative analysis of medical images. Many image labeling algorithms require seed identification in order to initialize segmentation algorithms such as region growing, graph cuts, and the random walker. Seeds are usually placed manually by human raters, which makes these algorithms semi-automatic and can be prohibitive for very large datasets. In this paper an automatic algorithm for placing seeds using multi-atlas registration and statistical fusion is proposed. Atlases containing the centers of mass of a collection of neuroanatomical objects are deformably registered in a training set to determine where these centers of mass go after labels transformed by registration. The biases of these transformations are determined and incorporated in a continuous form of Simultaneous Truth And Performance Level Estimation (STAPLE) fusion, thereby improving the estimates (on average) over a single registration strategy that does not incorporate bias or fusion. We evaluate this technique using real 3D brain MR image atlases and demonstrate its efficacy on correcting the data bias and reducing the fusion error.
Labeling; seed; continuous; STAPLE; atlas; statistics; fusion; bias; registration
A new algorithm is proposed for reconstructing raw RF data into ultrasound images. Prior delay-and-sum beamforming reconstruction algorithms are essentially one-dimensional, as a sum is performed across all receiving elements. In contrast, the present approach is two-dimensional, potentially allowing any time point from any receiving element to contribute to any pixel location. Computer-intensive matrix inversions are performed once-and-for-all ahead of time, to create a reconstruction matrix that can be reused indefinitely for a given probe and imaging geometry. Individual images are generated through a single matrix multiplication with the raw RF data, without any need for separate envelope detection or gridding steps. Raw RF datasets were acquired using a commercially available digital ultrasound engine for three imaging geometries: a 64-element array with a rectangular field-of-view (FOV), the same probe with a sector-shaped FOV, and a 128-element array with rectangular FOV. The acquired data were reconstructed using our proposed method and a delay-and-sum beamforming algorithm, for comparison purposes. Point-spread-function (PSF) measurements from metal wires in a water bath showed the proposed method able to reduce the size of the PSF and/or its spatial integral by about 20 to 38%. Images from a commercially available quality-assurance phantom featured greater spatial resolution and/or contrast when reconstructed with the proposed approach.
Over the past decade, dramatic increases in computational power and improvement in image analysis algorithms have allowed the development of powerful computer-assisted analytical approaches to radiological data. With the recent advent of whole slide digital scanners, tissue histopathology slides can now be digitized and stored in digital image form. Consequently, digitized tissue histopathology has now become amenable to the application of computerized image analysis and machine learning techniques. Analogous to the role of computer-assisted diagnosis (CAD) algorithms in medical imaging to complement the opinion of a radiologist, CAD algorithms have begun to be developed for disease detection, diagnosis, and prognosis prediction to complement to the opinion of the pathologist. In this paper, we review the recent state of the art CAD technology for digitized histopathology. This paper also briefly describes the development and application of novel image analysis technology for a few specific histopathology related problems being pursued in the United States and Europe.
histopathology; image analysis; computer-assisted interpretation; microscopy analysis
For quantitative analysis of histopathological images, such as the lymphoma grading systems, quantification of features is usually carried out on single cells before categorizing them by classification algorithms. To this end, we propose an integrated framework consisting of a novel supervised cell-image segmentation algorithm and a new touching-cell splitting method.
For the segmentation part, we segment the cell regions from the other areas by classifying the image pixels into either cell or extra-cellular category. Instead of using pixel color intensities, the color-texture extracted at the local neighborhood of each pixel is utilized as the input to our classification algorithm. The color-texture at each pixel is extracted by local Fourier transform (LFT) from a new color space, the most discriminant color space (MDC). The MDC color space is optimized to be a linear combination of the original RGB color space so that the extracted LFT texture features in the MDC color space can achieve most discrimination in terms of classification (segmentation) performance. To speed up the texture feature extraction process, we develop an efficient LFT extraction algorithm based on image shifting and image integral.
For the splitting part, given a connected component of the segmentation map, we initially differentiate whether it is a touching-cell clump or a single non-touching cell. The differentiation is mainly based on the distance between the most likely radial-symmetry center and the geometrical center of the connected component. The boundaries of touching-cell clumps are smoothed out by Fourier shape descriptor before carrying out an iterative, concave-point and radial-symmetry based splitting algorithm.
To test the validity, effectiveness and efficiency of the framework, it is applied to follicular lymphoma pathological images, which exhibit complex background and extracellular texture with non-uniform illumination condition. For comparison purposes, the results of the proposed segmentation algorithm are evaluated against the outputs of Superpixel, Graph-Cut, Mean-shift, and two state-of-the-art pathological image segmentation methods using ground-truth that was established by manual segmentation of cells in the original images. Our segmentation algorithm achieves better results than the other compared methods. The results of splitting are evaluated in terms of under-splitting, over-splitting, and encroachment errors. By summing up the three types of errors, we achieve a total error rate of 5.25% per image.
Histopathological image segmentation; touching-cell splitting; supervised learning; color-texture feature extraction; local fourier transform; discriminant analysis; radial-symmetry point; follicular lymphoma
DNA microarrays have emerged as a viable platform for detection of pathogenic organisms in clinical and environmental samples. These microbial detection arrays occupy a middle ground between low cost, narrowly focused assays such as multiplex PCR and more expensive, broad-spectrum technologies like high-throughput sequencing. While pathogen detection arrays have been used primarily in a research context, several groups are aggressively working to develop arrays for clinical diagnostics, food safety testing, environmental monitoring and biodefense. Statistical algorithms that can analyze data from microbial detection arrays and provide easily interpretable results are absolutely required in order for these efforts to succeed. In this article, we will review the most promising array designs and analysis algorithms that have been developed to date, comparing their strengths and weaknesses for pathogen detection and discovery.
microarrays; pathogens; genomics
There is critical need for improved biomarker assessment platforms which integrate traditional pathological parameters (TNM stage, grade and ER/PR/HER2 status) with molecular profiling, to better define prognostic subgroups or systemic treatment response. One roadblock is the lack of semi-quantitative methods which reliably measure biomarker expression. Our study assesses reliability of automated immunohistochemistry (IHC) scoring compared to manual scoring of five selected biomarkers in a tissue microarray (TMA) of 63 human breast cancer cases, and correlates these markers with clinico-pathological data. TMA slides were scanned into an Ariol Imaging System, and histologic (H) scores (% positive tumor area x staining intensity 0–3) were calculated using trained algorithms. H scores for all five biomarkers concurred with pathologists’ scores, based on Pearson correlation coefficients (0.80–0.90) for continuous data and Kappa statistics (0.55–0.92) for positive vs. negative stain. Using continuous data, significant association of pERK expression with absence of LVI (p = 0.005) and lymph node negativity (p = 0.002) was observed. p53 over-expression, characteristic of dysfunctional p53 in cancer, and Ki67 were associated with high grade (p = 0.032 and 0.0007, respectively). Cyclin D1 correlated inversely with ER/PR/HER2-ve (triple negative) tumors (p = 0.0002). Thus automated quantitation of immunostaining concurs with pathologists’ scoring, and provides meaningful associations with clinico-pathological data.
breast cancer; p53/cyclin D1/Ki67/pERK; tissue microarray; automated image analysis; clinico-pathological parameters
We describe an image analysis supervised learning algorithm that can automatically classify galaxy images. The algorithm is first trained using a manually classified images of elliptical, spiral, and edge-on galaxies. A large set of image features is extracted from each image, and the most informative features are selected using Fisher scores. Test images can then be classified using a simple Weighted Nearest Neighbor rule such that the Fisher scores are used as the feature weights. Experimental results show that galaxy images from Galaxy Zoo can be classified automatically to spiral, elliptical and edge-on galaxies with accuracy of ~90% compared to classifications carried out by the author. Full compilable source code of the algorithm is available for free download, and its general-purpose nature makes it suitable for other uses that involve automatic image analysis of celestial objects.
Methods; data analysis – Techniques; image processing
Spectral mixing is a problem inherent to remote sensing data and results in few image pixel spectra representing ″pure″ targets. Linear spectral mixture analysis is designed to address this problem and it assumes that the pixel-to-pixel variability in a scene results from varying proportions of spectral endmembers. In this paper we present a different endmember-search algorithm called the Successive Projection Algorithm (SPA). SPA builds on convex geometry and orthogonal projection common to other endmember search algorithms by including a constraint on the spatial adjacency of endmember candidate pixels. Consequently it can reduce the susceptibility to outlier pixels and generates realistic endmembers.This is demonstrated using two case studies (AVIRIS Cuprite cube and Probe-1 imagery for Baffin Island) where image endmembers can be validated with ground truth data. The SPA algorithm extracts endmembers from hyperspectral data without having to reduce the data dimensionality. It uses the spectral angle (alike IEA) and the spatial adjacency of pixels in the image to constrain the selection of candidate pixels representing an endmember. We designed SPA based on the observation that many targets have spatial continuity (e.g. bedrock lithologies) in imagery and thus a spatial constraint would be beneficial in the endmember search. An additional product of the SPA is data describing the change of the simplex volume ratio between successive iterations during the endmember extraction. It illustrates the influence of a new endmember on the data structure, and provides information on the convergence of the algorithm. It can provide a general guideline to constrain the total number of endmembers in a search.
hyperspectral; spectral unmixing; endmember; simplex
Summary: High-throughput DNA sequencing technologies have spurred the development of numerous novel methods for genome assembly. With few exceptions, these algorithms are heuristic and require one or more parameters to be manually set by the user. One approach to parameter tuning involves assembling data from an organism with an available high-quality reference genome, and measuring assembly accuracy using some metrics.
We developed a system to measure assembly quality under several scoring metrics, and to compare assembly quality across a variety of assemblers, sequence data types, and parameter choices. When used in conjunction with training data such as a high-quality reference genome and sequence reads from the same organism, our program can be used to manually identify an optimal sequencing and assembly strategy for de novo sequencing of related organisms.
Availability: GPL source code and a usage tutorial is at http://ngopt.googlecode.com
Supplementary information: Supplementary data is available at Bioinformatics online.
In vivo quantitative assessment of skin lesions is an important step in the
evaluation of skin condition. An objective measurement device can help as
a valuable tool for skin analysis. We propose an explorative new multispectral camera specifically developed for dermatology/cosmetology applications. The multispectral imaging system provides images of skin reflectance at different wavebands covering visible and near-infrared domain. It is coupled
with a neural network-based algorithm for the reconstruction of reflectance
cube of cutaneous data. This cube contains only skin optical reflectance
spectrum in each pixel of the bidimensional spatial information. The reflectance cube is analyzed by an algorithm based on a Kubelka-Munk model
combined with evolutionary algorithm. The technique allows quantitative
measure of cutaneous tissue and retrieves five skin parameter maps: melanin
concentration, epidermis/dermis thickness, haemoglobin concentration, and
the oxygenated hemoglobin. The results retrieved on healthy participants
by the algorithm are in good accordance with the data from the literature.
The usefulness of the developed technique was proved during two experiments: a clinical study based on vitiligo and melasma skin lesions and a skin oxygenation experiment (induced ischemia) with healthy participant where
normal tissues are recorded at normal state and when temporary ischemia is
We describe a method for fully automated detection of chemical synapses in serial electron microscopy images with highly anisotropic axial and lateral resolution, such as images taken on transmission electron microscopes. Our pipeline starts from classification of the pixels based on 3D pixel features, which is followed by segmentation with an Ising model MRF and another classification step, based on object-level features. Classifiers are learned on sparse user labels; a fully annotated data subvolume is not required for training. The algorithm was validated on a set of 238 synapses in 20 serial 7197×7351 pixel images (4.5×4.5×45 nm resolution) of mouse visual cortex, manually labeled by three independent human annotators and additionally re-verified by an expert neuroscientist. The error rate of the algorithm (12% false negative, 7% false positive detections) is better than state-of-the-art, even though, unlike the state-of-the-art method, our algorithm does not require a prior segmentation of the image volume into cells. The software is based on the ilastik learning and segmentation toolkit and the vigra image processing library and is freely available on our website, along with the test data and gold standard annotations (http://www.ilastik.org/synapse-detection/sstem).
Identification of prostatic calculi is an important basis for determining the tissue origin. Computation-assistant diagnosis of prostatic calculi may have promising potential but is currently still less studied. We studied the extraction of prostatic lumina and automated recognition for calculus images. Extraction of lumina from prostate histology images was based on local entropy and Otsu threshold recognition using PCA-SVM and based on the texture features of prostatic calculus. The SVM classifier showed an average time 0.1432 second, an average training accuracy of 100%, an average test accuracy of 93.12%, a sensitivity of 87.74%, and a specificity of 94.82%. We concluded that the algorithm, based on texture features and PCA-SVM, can recognize the concentric structure and visualized features easily. Therefore, this method is effective for the automated recognition of prostatic calculi.
We describe a protocol for fully automated detection and segmentation of asymmetric, presumed excitatory, synapses in serial electron microscopy images of the adult mammalian cerebral cortex, taken with the focused ion beam, scanning electron microscope (FIB/SEM). The procedure is based on interactive machine learning and only requires a few labeled synapses for training. The statistical learning is performed on geometrical features of 3D neighborhoods of each voxel and can fully exploit the high z-resolution of the data. On a quantitative validation dataset of 111 synapses in 409 images of 1948×1342 pixels with manual annotations by three independent experts the error rate of the algorithm was found to be comparable to that of the experts (0.92 recall at 0.89 precision). Our software offers a convenient interface for labeling the training data and the possibility to visualize and proofread the results in 3D. The source code, the test dataset and the ground truth annotation are freely available on the website http://www.ilastik.org/synapse-detection.
Image segmentation is important with applications to several problems in biology and medicine. While extensively researched, generally, current segmentation methods perform adequately in the applications for which they were designed, but often require extensive modifications or calibrations before being used in a different application. We describe an approach that, with few modifications, can be used in a variety of image segmentation problems. The approach is based on a supervised learning strategy that utilizes intensity neighborhoods to assign each pixel in a test image its correct class based on training data. We describe methods for modeling rotations and variations in scales as well as a subset selection for training the classifiers. We show that the performance of our approach in tissue segmentation tasks in magnetic resonance and histopathology microscopy images, as well as nuclei segmentation from fluorescence microscopy images, is similar to or better than several algorithms specifically designed for each of these applications.
This report outlines a neuroimaging pipeline that allows a robust, high-throughput, semi-automated, template-based protocol for segmenting the hippocampus in rhesus macaque (Macaca mulatta) monkeys ranging from 1 week to 260 weeks of age. The semiautomated component of this approach minimizes user effort while concurrently maximizing the benefit of human expertise by requiring as few as 10 landmarks to be placed on images of each hippocampus to guide registration. Any systematic errors in the normalization process are corrected using a machine-learning algorithm that has been trained by comparing manual and automated segmentations to identify systematic errors. These methods result in high spatial overlap and reliability when compared with the results of manual tracing protocols. They also dramatically reduce the time to acquire data, an important consideration in large-scale neuroradiological studies involving hundreds of MRI scans. Importantly, other than the initial generation of the unbiased template, this approach requires only modest neuroanatomical training. It has been validated for high-throughput studies of rhesus macaque hippocampal anatomy across a broad age range.
The presence of an atypical (irregular) pigment network (APN) can indicate a diagnosis of melanoma. This study sought to analyze the APN with texture measures.
For 106 dermoscopy images including 28 melanomas and 78 benign dysplastic nevi, the areas of APN were selected manually. Ten texture measures in the CVIPtools image analysis system were applied.
Of the 10 texture measures used, correlation average provided the highest discrimination accuracy, an average of 95.4%. Discrimination of melanomas was optimal at a pixel distance of 20 for the 768 × 512 images, consistent with a melanocytic lesion texel size estimate of 4–5 texels per mm.
Texture analysis, in particular correlation average at an optimized pixel spacing, may afford automatic detection of an irregular pigment network in early malignant melanoma.
melanoma; image analysis; texture; texel; pigment network; dermoscopy
Surgical 5/6 nephrectomy and adenine-induced kidney failure in rats are frequently used models of progressive renal failure. In both models, rats develop significant morphological changes in the kidneys and quantification of these changes can be used to measure the efficacy of prophylactic or therapeutic approaches. In this study, the Aperio Genie Pattern Recognition technology, along with the Positive Pixel Count, Nuclear and Rare Event algorithms were used to quantify histological changes in both rat renal failure models.
Analysis was performed on digitized slides of whole kidney sagittal sections stained with either hematoxylin and eosin or immunohistochemistry with an anti-nestin antibody to identify glomeruli, regenerating tubular epithelium, and tubulointerstitial myofibroblasts. An anti-polymorphonuclear neutrophil (PMN) antibody was also used to investigate neutrophil tissue infiltration.
Image analysis allowed for rapid and accurate quantification of relevant histopathologic changes such as increased cellularity and expansion of glomeruli, renal tubular dilatation, and degeneration, tissue inflammation, and mineral aggregation. The algorithms provided reliable and consistent results in both control and experimental groups and presented a quantifiable degree of damage associated with each model.
These algorithms represent useful tools for the uniform and reproducible characterization of common histomorphologic features of renal injury in rats.
5/6 nephrectomy; adenine kidney model; digital pathology; digital slide; Genie; image analysis; rodent renal failure model
Motivation: High-throughput sequencing technologies place ever increasing demands on existing algorithms for sequence analysis. Algorithms for computing maximal exact matches (MEMs) between sequences appear in two contexts where high-throughput sequencing will vastly increase the volume of sequence data: (i) seeding alignments of high-throughput reads for genome assembly and (ii) designating anchor points for genome–genome comparisons.
Results: We introduce a new algorithm for finding MEMs. The algorithm leverages a sparse suffix array (SA), a text index that stores every K-th position of the text. In contrast to a full text index that stores every position of the text, a sparse SA occupies much less memory. Even though we use a sparse index, the output of our algorithm is the same as a full text index algorithm as long as the space between the indexed suffixes is not greater than a minimum length of a MEM. By relying on partial matches and additional text scanning between indexed positions, the algorithm trades memory for extra computation. The reduced memory usage makes it possible to determine MEMs between significantly longer sequences.
Availability: Source code for the algorithm is available under a BSD open source license at http://compbio.cs.princeton.edu/mems. The implementation can serve as a drop-in replacement for the MEMs algorithm in MUMmer 3.
Supplementary information: Supplementary data are available at Bioinformatics online.
The use of digital imaging and algorithm-assisted identification of regions of interest is revolutionizing the practice of anatomic pathology. Currently automated methods for extracting the tumour regions in basal cell carcinomas are lacking. In this manuscript a colour-deconvolution based tumour extraction algorithm is presented.
Haematoxylin and eosin stained basal cell carcinoma histology slides were digitized and analyzed using the open source image analysis program ImageJ. The pixels belonging to tumours were identified by the algorithm, and the performance of the algorithm was evaluated by comparing the pixels identified as malignant with a manually determined dataset.
The algorithm achieved superior results with the nodular tumour subtype. Pre-processing using colour deconvolution resulted in a slight decrease in sensitivity, but a significant increase in specificity. The overall sensitivity and specificity of the algorithm was 91.0% and 86.4% respectively, resulting in a positive predictive value of 63.3% and a negative predictive value of 94.2%
The proposed image analysis algorithm demonstrates the feasibility of automatically extracting tumour regions from digitized basal cell carcinoma histology slides. The proposed algorithm may be adaptable to other stain combinations and tumour types.
High throughput microarray analysis has great potential in scientific research, disease diagnosis, and drug discovery. A major hurdle toward high throughput microarray analysis is the time and effort needed to accurately locate gene spots in microarray images. An automatic microarray image processor will allow accurate and efficient determination of spot locations and sizes so that gene expression information can be reliably extracted in a high throughput manner. Current microarray image processing tools require intensive manual operations in addition to the input of grid parameters to correctly and accurately identify gene spots. This work developed a method, herein called auto-spot, to automate the spot identification process. Through a series of correlation and convolution operations, as well as pixel manipulations, this method makes spot identification an automatic and accurate process. Testing with real microarray images has demonstrated that this method is capable of automatically extracting subgrids from microarray images and determining spot locations and sizes within each subgrid, regardless of variations in array patterns and background noises. With this method, we are one step closer to the goal of high throughput microarray analysis.
Analyzing the features of the chromosomes can be very useful for diagnosis of many genetic disorders or prediction of possible abnormalities that may occur in the future generations. For this purpose, karyotype is often used, for which to be made, it is necessary to identify each one of the 24 chromosomes from the microscopic images. Definition and extraction of the morphological and band pattern-based features for each chromosome is the first step to identify them. Centromere location is an important morphological feature. In this paper, a novel algorithm for centromere localization is presented. The procedure is based on the calculation and analyzing the concavity degree of the chromosome's boundary pixels. In this method, the centerline of the chromosome is computed and the score of each pixel on the centerline is considered as the sum of the concavity degree of two pixels on the chromosome's boundary that are perpendicular to it. Finally, location of the centromere is estimated as one pixel on the centerline which is corresponding to the maximum score. When applied the proposed algorithm on 50 images, an average error of 2.25 pixels for centromere localization is achieved.
Centerline; chromosome centromere; concave points; karyotyping; polynomial fitting
A programmable vision chip with variable resolution and row-pixel-mixed parallel image processors is presented. The chip consists of a CMOS sensor array, with row-parallel 6-bit Algorithmic ADCs, row-parallel gray-scale image processors, pixel-parallel SIMD Processing Element (PE) array, and instruction controller. The resolution of the image in the chip is variable: high resolution for a focused area and low resolution for general view. It implements gray-scale and binary mathematical morphology algorithms in series to carry out low-level and mid-level image processing and sends out features of the image for various applications. It can perform image processing at over 1,000 frames/s (fps). A prototype chip with 64 × 64 pixels resolution and 6-bit gray-scale image is fabricated in 0.18 μm Standard CMOS process. The area size of chip is 1.5 mm × 3.5 mm. Each pixel size is 9.5 μm × 9.5 μm and each processing element size is 23 μm × 29 μm. The experiment results demonstrate that the chip can perform low-level and mid-level image processing and it can be applied in the real-time vision applications, such as high speed target tracking.
vision chip; image processing; machine vision; mathematical morphology