Krishnan Nallaperumal

Image refinement using skew angle detection and correction for Indian license plates

Automatic License Plate Recognition system (ALPR) is essential to implement law enforcement and traffic control on transportation systems. ALPR systems consist of the tasks: license plate localization, character segmentation, and character recognition; due to the positioning of the vehicles, the localized license plate images are mostly skewed which need to be de-skewed before proceeding to character segmentation. In this paper, a novel Skew Correction Algorithm is proposed focusing on boundary line that optimizes speed and accuracy by using the Hough transforms to get the skew corrected License plate image.

Fast tone mapping for high dynamic range images

We present a fast, effective and flexible tone reproduction method that preserves visibility and contrast impression of high dynamic range scenes in low dynamic range reproduction devices. A single parameter controls the visibility and contrast in a simple and elegant manner and at interactive speed. The new method is simple to use and is computationally highly efficient. Experiments show that the technique produces good results on a variety of high dynamic range images. The method can also be used to enhance ordinary low dynamic range digital images.

Content Based Image Indexing and Retrieval Using Color Descriptor in Wavelet Domain

This project presents an approach to index and retrieve images using a compact color descriptor in two color spaces namely HSV,and CIE L*u*v*. A compact color descriptor adopted in the proposed content-based image retrieval system is 127-bit binary Haar color histogram, which is used as an index of the images in the database. The color histogram is obtained for an image using a suitable color space. Here, HSV color space, and CIE L*u*v*color space are chosen for comparison. A compact color descriptor is obtained from Haar transform coefficient of the color histogram. Each coefficient is quantized to a binary number, resulting in a 127-bit color descriptor. The descriptors of the images in the database are compared to the descriptor of the query image. The most similar images are retrieved and ordered according to their distances to the query. The proposed retrieval system can effectively retrieve the most similar images in CIEL*u*v* color space.

A Novel Adaptive Weighted Color Interpolation Algorithm for Single Sensor Digital Camera Images

A single sensor equipped with a color filter array (CFA) is used in many digital still cameras to capture any of the three primary color components, R(red), G(green) or B(blue) on each pixel location, in order to reduce the hardware cost. This single band is used to generate the true RGB image. Demosicing is the process of interpolating the missed colors, which aims to reconstruct the missing colors as close as possible by keeping the less computational complexity. In this paper a novel adaptive edge preserving, edge directed interpolating technique is proposed for color reproduction from Bayer mosaic images. The proposed algorithm aims to effectively estimate the missing G component in edge region and texture regions and the other color planes are estimated based on the interpolated value of the green plane. When the green plane is processed, for each missing green component in the CFA, the algorithm performs a gradient test, to identify edge direction and then carries out an interpolation along the direction of a smaller gradient to determine the missing green component. The variance of color differences can be used as a supplementary criterion to determine the interpolation direction for the green components. Depending on the sharpness, an adaptive weighted interpolation method is introduced. Experimental results show that the proposed method performs much better than other latest demosaicing techniques in terms of CPSNR, correlation coefficient and SSIM (structural-similarity).

New Efficient Image Compression Method for Single Sensor Digital Camera Images

Widely used digital cameras nowadays use single sensor color filter array (CFA) which captures only one component of a pixel among the three components red, green and blue (RGB). This interleaved RGB valued images are called mosaic images. The mosaic images are of one third of the RGB images size. The mosaic image can be compressed further to reduce storage. In this paper we propose a new efficient method of lossless compression for color mosaic images. First the mosaic image is transformed by 5/3 forward wavelet transforms, which is best suited for color mosaic data. We also proposed a low complexity adaptive context-based modified Golomb-Rice coding technique to compress the coefficients of the Transform. The lossless compression performance of the proposed method on color mosaic images is arguably the best so far among the existing lossless image codecs.

Study and analysis of hallucination in HDR images

This work presents a study and analysis of Hallucination in HDR Images it presents a technique for the high dynamic range image in over-exposed and underexposed regions, the hallucination image become blur and hence a bilateral filter algorithm is applied to smooth the image and preserve edges more effectively after applying the tone mapping algorithm to map these high dynamic images to produce a high dynamic range image will contents intact, this mapping technique is known as hallucination technique for High dynamic range images.

Sonar scattering strength models — A review and analysis

A detailed study and comprehensive analysis of sonar scattering strength models is done in this paper. It is desirable to implement a range of bottom scattering strength models if one wants to have an option of grazing angle dependencies. It is inferred further that the Gauss surface scattering strength model serves as a substitute for the Ogden-Erskine model and to provide the experimental evidence for choosing one scattering strength kernel over another at each operational site, one can use the Ainslie angle/decay rate relationship.

An analysis on sonar reverberation models

A comprehensive analysis of sonar reverberation modeling with respect to Bellhop Compatibility for Operational Applications is done in this work. This work focuses on analyzing reverberation models which are designed to be operational oriented. Such models are compatible with the Bellhop propagation model which can be used within the Environment Modeling Manager-EMM. In order to compliment Bellhop model, the Ellis normal mode model and the Harrison closed form expressions could be added to the EMM.

A rank filter method for localizing non-uniform, 2-column Indian license plate with unequal plate and character dimensions

The first and the most important stage for any ALPR system is the correct localization of the license plate within the image captured by the LPR camera. Most of the varieties of techniques reported for localization of license plate are applicable for a very controlled environment. This work focuses on localizing the license plate regions from true color still snapshots captured in a very realistic situation. The technique is based on a novel multi-stage approach for analysis of vertical edge gradients from contrast stretched gray-scale images. Horizontally and vertically oriented rank filters are used to detect clusters of high density of bright edges in the area of the number plate. The proposed approach embeds multi resolution and multi scale edge detection, adaptive searching, color analysis, and affine rectification in a hierarchical frame work for sign detection, with different emphasis at each phase to handle the text in different sizes, orientations, color distributions and backgrounds.

A novel edge based detection and PANDA framework approach for Indian license plate character recognition

Recognizing ASCII characters to reconstruct the license plate number remains a challenge particularly for Indian License Plates due to the heterogeneity of issues involved with Indian license plates. A character recognition algorithm is presented here based on combination of edge detection and PANDA framework; the edge based approach is used to find the shape of the each character in the license plate. The PANDA framework is a generic one used for the comparison of the two patterns. If the pattern is not exist in the data set it adds the pattern into the data set for future reference. By doing this, the recognition rate gets improved considerably. From the experimental results it is shown that the proposed method gives fairly acceptable level of accuracy for practical deployment of any Indian ALPR system.