Masoud Afrakhteh

Seoul National University of Science and Technology, Samsung Electronics SNU Cooperation R&D center, Post-Doc

Followers

Following

Co-author

Public Views

As a PhD holder in SW engineering with over 15 years of official experience in software development, verification , and research in seamless integration of embedded infotainment SW solutions in either industry or automotive sections. These solutions include ADAS solutions like computer vision and radar sensors fusion as well as in-vehicle infotainment features or portable embedded modules.

I have experienced all life cycles of a SW from overseeing the design to build, and execution of pre-existing or new infrastructure, tools, and processes for both stability and functionality of the solutions.

I have been expertised in image processing techniques via watermarking/steganography, techniques for embedding hidden information into digital media, to enhance the security and reliability of the SW solutions and worked on different embedded system platforms such as AVR, Arduino, Nvidia GPU enabled or Cypress Infineon boards.

My mission is to deliver high-quality products that meet the expectations in the shortest time possible with highest efficieny for stakeholders satisfying the needs of the end-users, as well as the standards and requirements of the industry.

less

Interests

Uploads

Papers by Masoud Afrakhteh

Steganography Using More Surronding Pixels

—Nowadays among the steganography techniques and particularly in conventional least significant b... more —Nowadays among the steganography techniques and particularly in conventional least significant bit (LSB) insertion method, there is a challenging issue and that is how to embed the secret bits in a medium like a typical 8-bit gray scale image in a way to be hidden to the human vision system. The gray scale image is called Cover Image and the pixels that carry the secret bits are called Target Pixels. The considerable point is how the capacity of every target pixel is achieved in order to maintain an acceptable imperceptibility of the secret data. The number of bits embedded in each target pixel is called Capacity. Some methods utilize either three or four adjacent neighbors of a target pixel so as to find its capacity such as BPCS, PVD and MBNS. In this paper, a method is proposed that uses at least four numbers of eight surrounding pixels of a target pixel. The more pixels are used for estimating the capacity, the higher image quality is achieved and vice versa. Thanks to this fact, smaller image's distortion is made in the cover image. The method is called MSPU that stands for more surrounding pixels using.

Download

Printed Document Authentication Using Watermarking Technique

—Forgeries related with official printed documents can easily be performed with the aid of today'... more —Forgeries related with official printed documents can easily be performed with the aid of today's advanced electronic devices such as scanners and computers. The forged documents are usually undetectable by human eyes and it is with this regard that there is an urgent need to find solutions to the threat of counterfeiting of such documents. With watermarking technique, information that is used to determine the validity of printed document can be embedded in the document. The embedded information can be imperceptible to human eyes, thus the process of forgery is made harder for the attackers. To authenticate the owner of the document, the embedded watermark is extracted from the watermarked document. However, in the verification process the printed document may suffer from printing and scanning (PS) distortion and as such it is necessary to resolve the noise and unwanted rotation as well as any degradation made by printing and scanning. This study proposed a watermarking technique to address this issue.

Download

False Positives removal for pedestrian object detection

In this paper, we investigate pedestrian detection in two ways. We proposed two simple ways to el... more In this paper, we investigate pedestrian detection in two ways. We proposed two simple ways to eliminate many of such unwanted FPs that usually result from applying a state-of-the-art pedestrian object detector to a visible color image. First approach only uses visible color imagery and reconstructs some new symmetric-looking objects from the first individual detected object. If any of these objects are detected, they are confirmed as a person; otherwise they are omitted from the very first list of detected objects .The second approach utilizes Color-Thermal Image pairs and applies an image categorization method for day and night images. As a result, a night time pre-trained pedestrian detector is only applied to night images, and a daytime one is applied to daytime images. Experimental results show how these FPs were removed in each case with such simple techniques while maintaining the miss rate as low as before.

Download

Enhanced Least Significant Bit Scheme Robust Against Chi-Squared Attack

Among the steganographic techniques and particularly in conventional least significant bit (LSB) ... more Among the steganographic techniques and particularly in conventional least significant bit (LSB) insertion method, there is a challenging issue and that is how to embed desired secret bits in a cover medium (a typical 8-bit gray-scale image) in a way not to be seen by human vision system as well as the fact that it is expected not to be attacked by some attacks like chi-squared attack, etc. The point is how to maintain robustness and the highest acceptable imperceptibility. Although by using more pixels for estimating each target pixel's capacity, bigger imperceptibility is achieved, there is another problem that the higher probability of being attacked by Chi-squared index is expected as the desired amount of secret bits increases. This paper proposes a method that utilizes more surrounding pixels (unlike BPCS, PVD and MBNS methods which use 3 or 4 immediate neighbors of each pixel). In this regard, for each certain target pixel, a more precise number of bits known as capacity, is found to be filled by respective secret bits. Finally, it is proved that the method is robust against Chi-squared attack. The new method is called MSP and it stands for more surrounding pixels.

Download

Double Phase Pedestrian Detection with Minimal Number of False Positives per Image

by Masoud Afrakhteh and MiRyong Park

In this paper, we investigate pedestrian detection from visible color imagery in a manner that re... more In this paper, we investigate pedestrian detection from visible color imagery in a manner that reduces the number of false positives (FPs) per frame. We propose a simple way to eliminate many of such unwanted FPs that usually result from applying a state-of-the-art pedestrian object detector to a visible color image. The symmetric structure of pedestrian bodies can often be a good clue to distinguish them from non-pedestrians. Hence, each and every individual detected object (bounded box) in the initial step of detection is used to reconstruct some new symmetric-looking objects. If any of these objects are detected, they are confirmed to be a person; otherwise, they are omitted from the very first list of detected objects. Experimental results show how these FPs are removed using such a simple technique while maintaining the previous low miss rate.

format_quoteSymmetric body parts effectively differentiate true positives from false positives after initial detection, boosting pedestrian recognition accuracy.format_quote

Download

Adaptive Watermarking for Printed Document Authentication

The rapid growth in information and communication technologies provides ways for some people to f... more The rapid growth in information and communication technologies provides ways for some people to forge printed documents. Digital watermarking is widely used for authenticating digital documents. By embedding some watermarks in the printed documents in a way not to be observed by human eyes we may stop these forgeries. While a printed document requires another procedure to extract the embedded watermark and this process normally suffers from image distortion which is made by either printer's quality or scanner's precision. In this paper, the proposed watermarking scheme is capable of authenticating the embedded watermark in the printed documents. There is a factor to regulate the amount of watermark to be embedded in each specified partition of the image known as regulation factor. In terms of security, the embedding process employs different values of regulation factor. As the watermarking scheme uses a matrix of regulation factors rather than a fixed regulation factor, it is considered as an adaptive embedding. On the other hand, once the watermarked document is printed, it is required to be scanned. The geometric distortion is required to be removed from the scanned document including watermark in order to improve the quality of the retrieved watermark so that the extracted watermark is normalized. By doing this, the extracted watermark is clearer for visual authentication.

Download

Double Phase Modular Steganography Using Error Images of Lossy Wavelet Transforms

An error image of Lossy Wavelet Transform differentiates a typical cover image from its wavelet t... more An error image of Lossy Wavelet Transform differentiates a typical cover image from its wavelet transformed cover image which is resulted from applying a desired image compression factor (ICF) to the respective cover image. The proposed method uses such error images for finding where and to what extent a target pixel in a cover image must hold secret bits. The method is also able to embed maximal embedding rates on gray-scale cover images, up to four bits per pixel (bpp), preserving a human vision-imperceptible PSNR value and less detectability. The detectability level is shown by Ensemble classifiers using second order SPAM as their input features. Additionally, the method embeds minimal rates, up to one bpp, while being roughly as undetectable as state-of-the-art schemes such as HUGO (highly undetectable stego) and EA (edge adaptive).

format_quoteLSB matching revisited improves randomness in pixel value modifications, enhancing steganographic methods.format_quote

Download

Adaptive Steganography Scheme Using More Surrounding Pixels

Information hiding has several branches. One is called as steganography. Steganography has a wide... more Information hiding has several branches. One is called as steganography. Steganography has a wide variety of techniques whereby a certain amount of secret information is embedded inside a medium. The medium format can be of any kind. However, the simplest file format is an 8-bit gray scale image. The original image is called a cover image. Likewise, once some secret bits are embedded, it is called stego-image. The most challenging issue is how capable a technique is to conceal the very existence of secret information in the stego-image. Steganographic techniques are evaluated in terms of imperceptibility and capacity. Imperceptibility means how visible secret information is to the human eyes in the stego-image. On the other hand, capacity means the maximum amount of secret information embedded. Conventional Least Significant Bit (LSB) insertion method is the basic concept for steganographic techniques. LSB method embeds secret bits in LSB(s) of the cover image. A pixel which carries a fraction of secret data is called a target pixel. The more surrounding pixels around a target pixel are utilized to find the most appropriate capacity value, the more precise capacity value is achieved. This paper proves that discovering the best capacity value brings about an improvement in terms of imperceptibility. Most of the steganographic techniques use either three or four adjacent pixels around a target pixel whereas the proposed technique is able to utilize at most all eight adjacent neighbors so that imperceptibility value grows bigger. The proposed method is called A-MSPU which stands for Adaptive More Surrounding Pixels Using.

Download

Double phase modular steganography with the help of error images

Many steganographic schemes based on popular least significant bit (LSB) embedding have been prop... more Many steganographic schemes based on popular least significant bit (LSB) embedding have been proposed. The embedding procedure is totally independent of pixel values due to the fact that only the LSB has to be altered. Other methods either randomly increment/ decrement the pixel value or use a function to do so. Because of this, the maximum embedding capacity cannot exceed 1 bit per pixel (bpp). State-of-the-art steganographic schemes, such as edge adaptive (EA) and highly undetectable stego (HUGO), aim at such LSB-based approaches with the help of LSB-matching algorithms and, unlike classical steganography, are more concerned about the undetectability level of the stego image rather than the peak signal-to-noise ratio (PSNR) of the stego images. There is a trade-off between detectability and the embedding rate of the stego images. The larger the embedding rate, the higher the chance of detectability, and the opposite is also true. On the other hand, classical steganographic approaches aim for maximum embedding capacity (say 4 bpp) while their main ability is to provide greater PSNR values rather than undetectability. In this work, a novel modular steganographic scheme is proposed in a spatial domain that takes advantage of error images resulting from applying an image quality factor (the same as the ones used in JPEG compression) in order to find the pixels where the secret data could be embedded. We show that our proposed method is less detectable compared to the EA method. The proposed method is almost as undetectable as HUGO. The detectability level is evaluated by the most recent state-of-the-art steganalysis attack, called ensemble classifiers, with second-order subtractive pixel adjacency model features given as their input. In addition, the proposed method can embed up to 4 bpp, yet maintain a high PSNR value.

format_quoteA novel steganographic scheme shows lower detectability compared to established methods like EA and HUGO using error images for embedding.format_quote

Download

Parallel modular steganography using error images

The embedding process for most of the state-of-the-art adaptive steganographic methods (not easil... more The embedding process for most of the state-of-the-art adaptive steganographic methods (not easily detectable) such as highly undetectable steganography (HUGO), edge adaptive image steganography (EA), and Adaptive ±1 Steganography in Extended Noisy Regions, is contingent upon the neighboring pixel values to prevent discovery by steganalysis methods. Hence, the calculation of the possible embedding rate must be suspended until the embedding process is performed pixel by pixel, in sequence. If, however, the cover image is JPEG-compressed, an error image can be considered as an embedding map implying where and to what extent secret bits can be embedded. In this paper, we present a modular steganography using error images. It is demonstrated that the detectability level of the resulting stego images is lower than that of state-of-the-art schemes and the proposed method executes 2.32 times faster than HUGO. Further, to improve the time performance , a block-wise parallel algorithm is presented where a single thread is assigned to each block of pixels. Because of the independent behavior of the embedding method and unlike the other methods, it is proved that each block of pixels can be embedded independently. The effectiveness of this parallel approach is analyzed and verified to execute approximately 55 times faster than serial execution.

Download

Adaptive least significant bit matching revisited with the help of error images

State-of-the-art steganographic schemes, such as highly undetectable stego (HUGO) and its extende... more State-of-the-art steganographic schemes, such as highly undetectable stego (HUGO) and its extended version, aim at least significant bit (LSB)-based approaches embedding up to 1 bpp and are more concerned about the undetectability level of the stego image rather than the peak signal-to-noise ratio. The complexity of such methods is quite high, too. In this work, a steganographic scheme is proposed in a spatial domain that takes advantage of error images resulting from applying an image quality factor (the same as the ones used in JPEG compression) in order to find the pixels where a slight change could be made. The amount of change is adaptively embedded using LSB matching revisited. We show that our proposed method is less detectable than HUGO and almost as undetectable as the extended HUGO while it has a greater time performance.

Download

Pedestrian Detection with Minimal False Positives per Color-Thermal Image

This research is based on aggregate channel features utilized for pedestrian detection, and the m... more This research is based on aggregate channel features utilized for pedestrian detection, and the main focus is to investigate a simple way to reduce the number of false positives per image. The importance of this will be to increase the accuracy of the detector by removing the excessive number of false positives while maintaining the missing rate as low as possible. To omit such unwanted false positives, we utilized an image categorization method for day and night images in order to minimize the misclassification rate. Furthermore, the best extension of the aggregate channel features method was analyzed and is recommended as a base detector. As a result, a night-time pre-trained pedestrian detector is only applied to night images, and a daytime detector is applied to daytime images. Thus, a large number of false positives are avoided while the missing rate is greatly reduced.

Download

Masoud Afrakhteh

Uploads

Papers by Masoud Afrakhteh

Log In