New Efficient Method for Coding Color Images

 Abstract — In this paper a novel color image compression technique for efficient storage and delivery of data is proposed. The proposed compression technique started by RGB to YCbCr color transformation process. Secondly, the canny edge detection method is used to classify the blocks into the edge and non-edge blocks. Each color component Y, Cb, and Cr compressed by discrete cosine transform (DCT) process, quantizing and coding step by step using adaptive arithmetic coding. Our technique is concerned with the compression ratio, bits per pixel and peak signal to noise ratio, and produce better results than JPEG and more recent published schemes (like CBDCT-CABS and MHC). The provided experimental results illustrate the proposed technique that is efficient and feasible in terms of compression ratio, bits per pixel and peak signal to noise ratio.


I. INTRODUCTION
N the recent years there has been an astronomical increase in the usage of computers for a variety of tasks.One of the most common usages has been the storage, manipulation, and transfer of digital images.The files that comprise these images, however, can be quite large and can quickly take up precious memory space on the computer's hard drive.In multimedia application, most of the images are in color.Color images contain a lot of data redundancy and require a large amount of storage space.
Image compression refers to the reduction of the size of the data that images contain.Image compression schemes in [1], [2] exploit certain data redundancies to convert the image to a smaller form.A typical image compression system is shown in Fig. 1.The data reduction, or compression, is performed by a device known as the encoder.The encoder reduces the data size of the original image A. The compressed image B is the output that passes through a channel (usually an actual transmission channel or a storage system) to the decoder.The decoder reconstructs, or decompresses, the image C from the compressed data.The ratio of the size (amount of data or bandwidth) of the original image to the size of the compressed image is known as the compression ratio or compression rate.The compression ratio can also be expressed in bpp (bits per pixel).The term bit rate is a general term for bpp.The higher the compression rate, the greater is the reduction of data [3], [4].Depending on the application, the channel may be affected by the noise that results in distortion of the compressed image during transmission.If so, the channel is known as an errorprone channel; otherwise, it is errorless.In Fig. 1, the channel Walaa M. Abd-Elhafiez and Wajeb Gharibi are with College of Computer Science & Information Systems, Jazan University, Jazan, Kingdom of Saudi Arabia.(e-mail: walaa.hussien@science.sohag.edu.eg; gharibiw @hotmail .com). is assumed to be error-free.Hence, B is the input to the decoder.Data compression schemes can be divided into two broad classes: 1. Lossless compression schemes [5], [6], in which C is identical to A. 2. Lossy compression schemes [7], [8], which provide much higher compression than lossless compression but allow C to be different from A. This paper is structured as follows: Section II reviews some related work.Our technique is described in Section 3. Experimental results are shown in Section IV. Conclusions are drawn in Section V.

II. RELATED WORK
Wang et.al [9] presented a cost effective block truncation coding (CE-BTC) using low-cost approach for color image compression.The usage of line buffer memory in low-cost approach CE-BTC is only a half of that in the optimal approach in CE-BTC.Therefore, the low-cost approach CE-BTC can be suited to apply to some restrained resource applications such as frame memory reduction in LC Displays (LCD) overdrive.Simulation results show that the proposed CE-BTC outperforms the VQ-BTC in PSNR up to 3 dB and much better subject visual quality.Sowmyan et.al [10] have been proposed several methods for color image compression but the reconstructed image have very low signal to noise ratio that made it inefficient.Their technique worked on the spatial domain where the pixel values of RGB planes of the input color image is mapped onto two-dimensional planes.Satish and Shishir [11] used various contemporary standards by Joint Picture Expert Group for compression.They exploited the correlation among the color components using a component color space transform before the subband transform stage.The transforms used to decorrelate the colors are primarily the fixed kernel transforms, which are not suitable for the large class of images.In their paper an image dependent color space transform (ID-CCT), exploiting the inter-channel redundancy optimally and which is very much suitable for compression proposed.Also, the comparative performance has been evaluated, and a significant improvement has been observed, objectively as well as subjectively over other quantifiable methods.

III. THE PROPOSED COLOR IMAGE COMPRESSION METHOD
Each color image consists of three components whether stored in RGB or YCbCr format.For this application, the input image is first converted into YCbCr format as in [12], because, in this way, additional decorrelation of the components (better compression) is done.Then classified the image into the background and foreground portions as describe in section 3.2.Then the image is subdivided into 8x8 blocks, and DCT coefficients are computed for each block.The quantization is performed conferring to predetermined quantization table.The quantized values are then rearranged according to zig-zag scan arrangement.The less important values are discarded (as describe in section 3.3) from the list in the zig-zag arrangement.After discarding insignificant coefficients, the remaining coefficients are compressed by the adaptive arithmetic coding (Q-coder).

A. Classification Step
Edges often occur at points where there is a large variation in the luminance values in the image, and consequently they often indicate the edges, or occluding boundaries, of the object in the scene.There are several techniques have been used for edge detection [13].In this paper, Canny Method is used.The canny edge detection algorithm is known to many as the optimal edge detector [14].The image is divided into two classes: edge blocks and non-edge blocks.

B. The Modification To increase compression ratio, in this part we suggested technique modification of JPEG compression by collection between JPEG compression technique and edge extraction.
The modification will be done after the quantization step.The image is subdivided into a block of pixels, and then these blocks are classified into the edge (significant regions/foreground) and non-edge (insignificant regions/background) blocks.In the first method (A-1), the non-edge blocks are compressed using the DC coefficient only and all significant coefficients are used for the edge blocks.In the second method (A-2) on each component (Y, Cb, and Cr), the DC coefficient only is used for coding the non-edge blocks (insignificant regions).70% (choose by experimental) of the non-zero quantized AC coefficients have been used in the coding of edge blocks.In the third method (A-3) on each component (Y, Cb, and Cr), the DC coefficient only is used for coding the non-edge blocks (insignificant regions).50% of the non-zero quantized AC coefficients have been used in the coding of edge blocks.

C. Q-Coder (Adaptive Arithmetic Code)
Adaptive arithmetic code [15] is a lossless compression technique that benefits from treating multiple symbols as a single data unit but at the same time retains the incremental symbol-by-symbol coding approach of Huffman coding.Arithmetic coding separates the coding from the modeling.This process allows for the dynamic adaptation of the probability model without affecting the design of the coder.Provisions for substituting Huffman coding for arithmetic coding are contained in many of the image compression standards.

IV. EXPERIMENTAL RESULTS
For the implementation and evaluation of the algorithms, we developed a MATLAB code and performed the testing on a standard color test images Lena, Fruit, and Airplane of size 512x512 and another test image Zelda and House of size 256x256 (Fig. 2).We analyze the results obtained with the first, second, and finally, the proposed algorithm.All images and tables from the experiments are given.Standard measures for image compression [16], like compression ratio (CR) and peak signal to noise ratio (PSNR) were used, which are calculated for comparing the performance of the proposed method as per the following representations: From the results listed in Table I, the proposed codec achieves a high performance, and we can conclude that about 0.1299-0.3404-bitrate reduction on average is achievable by using the proposed method (A-3).The subjective visual quality is compared using a different color image, as shown in Experiment results show that A-1 and A-2 perform better in PSNR compared with A-3.Although A-1 shows a litter better performance than A-2, the bit rate of A-1 is much higher than that of A-2.The CR results of the proposed method (A-3) and the other compared methods MHC scheme [17] are presented in Fig. 6.The A-3 obtains 35.33 on average and the best performance (41.0029) in House.Compared with CBDCT-CABS [18], the A-3 achieves 2.8334 dB higher average performances in PSNR, especially in Airplane the improvement is up to 4.15 dB.The A-3 also performs high better by 2.27 dB on average than the most competitive method of JPEG (as shown in Fig. 7).
Fig. 6 Compression ratios attained for test images with different block size of the proposed method (A-3) and MHC scheme [17]

Fig. 1
Fig. 1 Block diagram of image compression system Walaa M.Abd-Elhafiez, Wajeb Gharibi New Efficient Method for Coding Color Images I World Academy of Science, Engineering and Technology International Journal of Computer and Information Engineering Vol:8, No:2, 2014 403 International Scholarly and Scientific Research & Innovation 8(2) 2014 ISNI:0000000091950263 Open Science Index, Computer and Information Engineering Vol:8, No:2, 2014 waset.org/Publication/10001981

Algorithm 1 .
Input the image to be compressed.2. Classify the input image into background and foreground based on edges.3. Subdivide the input image into 8x8 blocks.4. Find the DCT coefficients for each block.5. Quantize the DCT coefficients based on predetermined quantization table.6. Apply the modification of the quantized coefficient based on the classification in step 2. 7. Assemble the blocks into a continuous stream.8. Compress the resulting values by apply Q-coder.