Low-level Image Processing Applications Research Articles

Fast and efficient implementation of image filtering using a side window convolutional neural network

Convolutional neural networks (CNNs) designed for object recognition have been successfully applied to low-level tasks such as image filtering. However, these networks are usually very large which occupy large memory space and demand very high computational capacity. This makes them unsuitable for real time low-level applications on smart and portable devices with limited memory and computational capacities. In this paper, we specifically design a novel CNN, side window convolutional neural network (SW-CNN), for the fast and efficient implementation of image filtering. In SW-CNN, a new convolutional strategy, called side kernel convolution (SKC) is proposed which aligns the side or corner of the convolutional window with the pixels under processing to preserve edges during convolution. By combining SKC and the representational power of CNNs, SW-CNN can learn various image-filtering tasks very effectively. Compared to the state-of-the-art networks, the superiority of SW-CNN includes three aspects. First, the number of learnable parameters is reduced by 96%. Second, the memory consumption is reduced to 50%. Third, the running time is decreased to 50%. Results of extensive experiments demonstrate that SW-CNN not only has good performance on implementing various edge-preserving filters, but also has the adaptability and flexibility on other low-level image processing applications.

An Improvement to a Class of Intensity based Spatial Domain Corner Detection Algorithms using Image Fission and Fusion

Corners of an object are important as features for the representation and analysis of its shape in computer vision. Corner detection, particularly in real scenes, is still a challenge. Most of the corner detectors found in the literature generate a number of false corners, which is not acceptable in real-life applications. In this paper, an improvement to a class of corner detection algorithms is presented using image fission/fusion. In this approach, a grayscale image is first divided into several bit-planes. A corner detector is applied on all the bit-planes simultaneously and a threshold (bitplane) is obtained using the concept of information gain. Finally, all the higher bit-plane corners are recombined (up to some thresholded bit-plane) to obtain the final set of corners. Here the corner detection algorithm is considered as a binary classification problem. Experimental results show that this improved approach reduces the number of erroneous corner detection relative to existing spatial domain corner detection algorithms. The improvements are established with the help of a number of performance measures proposed by various researchers. The proposed approach works better with respect to computational time also. This approach can easily be utilized in different low-level image processing applications.

Low level image processing and analysis using radius filters

A novel class of nonlinear L-filters based on local statistics is introduced, which will be called Radius Filters (RFs). The underlying idea for the construction of RFs is the sorting of the values of the input signal belonging in a sliding window, according to their distance from their mean value, in a vector Tr that carries significant signal information, while the filter output is defined as a linear combination of the elements of the vector Tr. RFs are simple, intuitive and easy to implement while they offer a unique signal formation, which provides insight and in depth signal interpretation for the proper filter design. Moreover, they may considered as a generalization of the OS Filters, since OS filters may be easily produced from the RFs. A number of low level image processing applications of the RFs such as impulse noise removal, speckle noise removal, adapting filtering, edge detection in noisy images, image sharpening and image frequency decomposition are outlined.

A general-purpose processor-per-pixel analog SIMD vision chip

A smart-sensor VLSI circuit suitable for focal-plane low-level image processing applications is presented. The architecture of the device is based on a fine-grain software-programmable SIMD processor array. Processing elements, integrated within each pixel of the imager, are implemented utilising a switched-current analog microprocessor concept. This allows the achievement of real-time image processing speeds with high efficiency in terms of silicon area and power dissipation. The prototype 21 /spl times/ 21 vision chip is fabricated in a 0.6 /spl mu/m CMOS technology and achieves a cell size of 98.6 /spl mu/m /spl times/ 98.6 /spl mu/m. It executes over 1.1 giga instructions per second (GIPS) while dissipating under 40 mW of power. The architecture, circuit design and experimental results are presented in this paper.

P-3PC: a point-to-point communication model for automatic and optimal decomposition of regular domain problems

One of the most fundamental problems automatic parallelization tools are confronted with is to find an optimal domain decomposition for a given application. For regular domain problems (such as simple matrix manipulations), this task may seem trivial. However, communication costs in message-passing programs often depend significantly on the memory layout of data blocks to be transmitted. As a consequence, straightforward domain decompositions may be non-optimal. In this paper, we introduce a new point-to-point communication model, called P-3PC (Parameterized model based on the Three Paths of Communication), that is specifically designed to overcome this problem. In comparison with related models (e.g. LogGP), P-3PC is similar in complexity, but more accurate in many situations. Although the model is aimed at MPI's standard point-to-point operations, it is applicable to similar message-passing definitions as well. The effectiveness of the model is tested in a framework for automatic parallelization of low-level image processing applications. Experiments are performed on two Beowulf-type systems, each having a different interconnection network and a different MPI implementation. The results show that, where other models frequently fail, P-3PC correctly predicts the communication costs related to any type of domain decomposition.

Morphological and wavelet transforms for object detection and image processing

We consider the problem of detecting multiple distorted objects in an input scene with clutter. The input scenes contain different types of background clutter and multiple objects in different classes, with different object aspect views, different object representations, hot/cold/bimodal/partial object variations, and high/low contrast object variations. Several new optical morphological operations for use in the above detection problem and in other general low-level image-processing applications are described, and several examples of their use are provided. For difficult detection problems in which high detection rates and low false-alarm rates are required we combine morphological operations and optical wavelet transforms to reduce clutter and improve object detection. The details of this set of filters and initial testresults are given. The most computationally demanding operations required in all cases are realizable on an optical correlator.

A high level language for parallel image processing

Most published research in the field of parallel image processing has tended to be in the areas of parallel architectures and parallel algorithms. Work on the development of software tools such as languages has generally been less extensive. This paper describes some research which is intended to redress the balance a little, by describing I-BOL-an application-specific high level programming language intended for implementing low-level image processing applications on parallel architectures. In particular, I-BOL has been designed to be capable of implementation on distributed memory parallel machines such as transputer networks. This paper introduces the core concepts of I-BOL: its view of an image as a set of tuples; user-defined neighbourhood functions; and I-BOL's facilities for recursive image processing. Solutions to a number of example problems illustrate particular aspects of the notation, including the Distance Transform, Histogram Equalization and the Hough Transform. Some consideration is given to the parallel aspects of the current implementation of I-BOL on a pipeline of transputers. A few performance measurements are quoted, giving execution times for the chosen examples on various sizes of transputer work.

An empirical study of automatic restructuring of vision programs for simd machines

The feasibility of Automatic restructuring low-level Image Processing applications for parallel processing is studied through experiments using an automatic restructurer developed by us. To measure performance we present parallel processing speedup results due to automatic restructuring for these applications. The effectiveness of scalar expansion and loop interchanging to increase the degree of vectorization is discussed. The applications are classified in terms of a small set of features. We analyze the applications in which automatic restructuring resulted in a “poor” speedup, identifying those application characteristics that have the greatest effect on it. The experiment suggests that automatic restructuring can be a useful tool for exploiting parallelism from the sequential form of low-level image processing applications.

IAL: a parallel image processing programming language

The paper describes a high-level programming language for image processing, called IAL (Image Algebra Language). IAL has been designed for low-level image processing applications. It has been implemented on transputer networks. IAL provides mechanisms whereby a programmer can specify image processing operations algebraically on the complete image. The implementation automatically parallelises the program, distributes the image data and controls all interprocess communication. The language is based on image algebra (a mathematical notation developed specifically for low-level image processing at the University of Florida). The paper describes the features of the language and provides an outline of some of the strategies used in implementing it on a transputer network.