An Efficient Color Image Encoding Scheme Based on Colorization (original) (raw)

Color Image Encoding Using Morphological Decolorization

Image colorization is a new image processing topic which refers to recolor gray images to look like the original color images as possible. Different methods appeared in the literature to solve this problem, the way which leads to thinking about decolorization which means eliminating the colors of color images to just small color keys, aid in the colorization process. Due to this idea, decolorization is considered as a color image encoding mechanism. In this paper we propose a new decolorization system depends on extracting the color seeds using morphology operations. Different decolorization methods was studied and compared to our system results using different quality metrics.

Morphological decolorization system for colors hiding

Colors Hiding refers to the process where the chromaticity values are processed to be hidden in the achromatic channel. This concept can be found in the literature as color protection. In this paper we propose a new color hiding system based on decolorization. Decolorization refers to eliminating the colors to just few color seeds, used in the colorization process. In this paper the proposed decolorization system depends on extracting the color seeds using morphology operations. The proposed Morphological Decolorization System (MDS) can extract very few seeds — compared to other methods — and results in very qualified colorization. The seeds then are hided in the luminance channel after encoding using Least Significant Bit (LSB) with very few bit planes. The results of the system show very high quality color retrieval with high chromatic compression ratio compared to other literature methods.

Efficient Segmentation Techniques for Optimized Colorization based Compression

We are proposing a new scheme for the colour image compression which uses the colour data from a couple of representative pixels to train a model that predicts colour of the remaining of the pixels. By storing the representative pixels, the image in grayscale is suffice to restore the first image. In this thesis, the colorization coding problem has been resolved using L1 norm minimisation sparse recovery algorithms (OMP). From the encoder, only a few representative pixels for the chrominance are sent together with brightness part of image to the decoder where the chrominance values for all the pixels are reconstructed by colorization technique. The main problem in colorization coding technique is to extract the RP efficiently to recreate the chrominance data of all the pixels for obtaining an excellent quality color image. In existing technique, the colorization matrix is created using the mean shift segmentation of brightness channel of the image. The C matrix extracts RP set by solving OMP. We proposes 2 additional kinds of segmentation schemes, multi scale k-means and multi scale super pixel. The quality of reconstructed image has been evaluated using the parameters-file size, MSE, PSNR and SSIM. Experiments shows that the proposed methods produce far better results, both qualitatively and quantitatively compared to existing strategies. Our methods also outperform standard colorization coding techniques as well as the JPEG standard. Also our proposed scheme is almost equally efficient compared to JPEG2000, in terms of the compression and also the quality of the retrieved colour image.

A Proposed Technique for Gray Image Colorization

The colorization of the gray-scale image is the process of using color image (has a similar "mood") to add color to a grayscale image. In this paper, we proposed new approach for colorization gray-scale images depending on Singular Value Decomposition (SVD). We used SVD to add color to a grayscale image by determining the best pixel value in the reference image, based on comparing the first column values from the left singular vector matrix of 3x3 block and the pixel value in the center of block of the gray image with other corresponding values in the reference image. Finally, use the best match to transfer color from color image to gray-scale image. The results of the proposed colorization approach are good and plausible.

Colour Image Compression With Colour Conversion and Hybrid Algorithm

The astounding augmentation of multimedia in the fields of communication media, medicine, surveillance etc. resulted in the huge volume of data acquirement. The storage of these data requires massive memory. For communication, these data need enormous transmission bandwidth. The only solution to reduce the storage and the transmission bandwidth is the data compression. From the literature survey it is learnt that there is a need to achieve compression ratio greater than 30 with a PSNR greater than 25 dB for non critical applications. In order to facilitate this, a colour image compression method is proposed. In this method, the colour image is converted into the "YCbCr" format using formulated New Equation Set-1. The "Y" component matrix is divided into 16×16 blocks. The DCT is applied to all the 16×16 blocks. The DC-Coefficient of all 16×16 block DCT is taken out and zero is inserted in place of it. The data types of all the DC-coefficients are changed from the "double" to the "16 bit integer" data type and they are stored. The transformed matrix consists of 16×16 block DCT of all the blocks. In this matrix, all those elements less than the threshold value "th" are made zero. This matrix is decomposed into matrices "U", "S" and "V" using SVD. All those elements of the matrix "U" less than the threshold value "thu" , all those elements of the matrix "S" less than the threshold value "ths" and all those elements of the matrix "V" less than the threshold value "thv" are made zero. Then these matrices are multiplied to form one matrix such that X=USV T. All those elements of the matrix "X" less than the threshold value "th" are made zero. Now all the elements of the matrix "X" are divided by 10. Then the matrix "X" becomes a sparse matrix. This sparse matrix is represented in the "triplet form". The data types of the "row values" and the "column values" of the triplet form are converted from the "double" to the "16 bit integer" data type. The data type of the "data elements" of the "triplet form" is converted into the "8 bit integer" data type. Then the RLE is applied to the "column values" of the "triplet form". After this, the compressed form of the Y-Component Matrix is obtained. Similarly, the "Cb" and the "Cr" component matrices are compressed. Then the experiments are conducted by converting the given image into the "YCbCr" format by the formulated New Equation Set-2, New Equation Set-3 and the basic "YCbCr" equation. The results are compared with parameters such as Compression Ratio, PSNR, SSIM and Quality Index. Experiments are conducted using MATLAB. From the results, it can be concluded that, the compression ratio obtained from the method which has got the colour conversion using New Equation Set-1 is good. The maximum compression ratio obtained with this method is 43.5079 with a PSNR of Red, Green and Blue Component equal to 25.9583 dB, 25.7501 dB and 26.4837 dB respectively. I. LITERATURE SURVEY There are different contributions to the above discussed problem. Few papers are discussed in this section. Raghevendra.M.J and others [1][2] have worked on image compression using DCT and SVD. Raghavendra.M.J and others [3] have worked on image compression using combinations of DCT, SVD and RLE. In this work, it is possible to achieve a compression ratio of 34.2325 with a PSNR of 25.2174 dB for a grayscale image. Raghavendra M.J and others [4] have worked on colour image compression. The paper [4] is in press. In this [4] Block wise operation is not done and the colour conversion operation is not done. In [4] the maximum compression ratio obtained is 32.1552 with a PSNR of around 24dB. Prasanta.H.S and others have worked on image compression using SVD [5]. In this a compression ratio of 4.12 with a PSNR of 43.85dB is obtained for the 32-Rank of the S-Matrix of the SVD. In this paper, an insight is given such that as we decrease the rank of the S-matrix of the SVD, the compression ratio increases. S.R.Subramanya and others have worked on wavelet transform [6] with predictive coding. Chandan .S.R and others have worked on compression of images using DCT and Fractal encoding [7]. Anna.

Literature Survey on Color Image Compression

2020

The need for an efficient technique for compression of Images ever increasing because the raw images need large amounts of disk space seems to be a big disadvantage during transmission & storage. Even though there are so many compression technique which is faster, memory efficient and simple surely suits the requirements of the user. This paper consists of review of some of the color image compression techniques.