Lin-bo Luo

A software-hardware cooperative implementation of bird's-eye view system for camera-on-vehicle

Rear camera or other wide-angle camera mounted on vehicle has serious perspective effect which makes driver unable to feel distance correctly and is not helpful for succeeding image analysis. To remove perspective effect, its necessary to change diagonal view to birds-eye view. In this paper, a software-hardware cooperative birds-eye view system is proposed, which can estimates the transformation matrix and changes the view automatically in real-time. The matrix estimation module is implemented by software, because it only needs to calculate once. The transformation part is implemented by hardware, because it must be done repetitively and it requires high computation power. To optimize hardware performance and save cost, three optimization approaches like one-step transformation, memory pre-estimation and look-up table (LUT) have been employed.

Improved LUT-Based Image Warping for Video Cameras

Image warping is usually used to perform real-time geometry transformation for video cameras. Several existing look-up table (LUT)-based algorithms achieve real-time performance, but the size of LUT is large and it has to be stored into off-chip memory. To reduce latency and bandwidth requirements due to using off-chip memory, this paper proposes an improved LUT (ILUT) scheme which compresses the LUT small enough that it can be stored in on-chip memory. Firstly, a one-stop transformation is adopted instead of several stages of on-line calculation. Then, the memory size of LUT is reduced by utilizing the similarity of neighbor coordinates and the symmetric characteristic of some image warping. Moreover, an elaborate pipeline hardware structure, cooperating with a novel 25-point interpolation algorithm, is proposed to accelerate the system and save further usage of memory. The proposed system is implemented by a field-programmable gate array-based platform. Two different examples show that the proposed ILUT achieves a real-time performance with small memory usage and low system requirements.

Low-cost implementation of bird's-eye view system for camera-on-vehicle

Many papers concentrate on the 3*3 perspective transformation matrix calculation of birds view system, but few discussed for the whole system implementation. In this paper, a low-cost birds-eye view system is proposed, which adopts an elaborate software/hardware cooperative system structure. It can be applied to all kinds of vision-based system in vehicles directly as a ready-made module.

Single bit plane based block truncation coding for color image compression in LCD overdrive

This paper presents an improved block truncation coding (BTC) for color image compression in LCD overdrive. To improve the BTC, single bit plane is used to represent the R, G and B bit planes and correct the two representative levels. Simulation results show that the proposed algorithm can significantly improve the compression ratio of the BTC from 4:1 to 6:1 with the limited performance losing about 0.297dB. Compared with other competitive improved BTC such as VQ-BTC, the proposed algorithm also performs 2.107dB better.

Real-Time Face Recognition Using Adaptive Skin-Color Model

In a real-time face recognition system, complex background that occurs in an image or video can degrade the accuracy of both face detection and recognition due to the uncontrolled environment. However, few studies have been proposed to address this issue. In this paper, we propose an adaptive skin color model to make the system more reliable. First, real face region is elected among face-like regions with skin color detection. Then, additional objects (i.e., background, hair and clothes) are omitted in face image by using a 2-D skin color model. It is shown that the speed of the algorithm proposed is almost the same as classical ones, while the recognition rate improves about 10% in crowd or blurry background.

Adaptive Object-Region-Based Image Pre-Processing for a Noise Removal Algorithm

A pre-processing system for adaptive noise removal is proposed based on the principle of identifying and filtering object regions and background regions. Human perception of images depends on bright, well-focused object regions; these regions can be treated with the best filters, while simpler filters can be applied to other regions to reduce overall computational complexity. In the proposed method, bright region segmentation is performed, followed by segmentation of object and background regions. Noise in dark, background, and object regions is then removed by the median, fast bilateral, and bilateral filters, respectively. Simulations show that the proposed algorithm is much faster than and performs nearly as well as the bilateral filter (which is considered a powerful noise removal algorithm); it reduces computation time by 19.4 % while reducing PSNR by only 1.57 % relative to bilateral filtering. Thus, the proposed algorithm remarkably reduces computation while maintaining accuracy.

Real-Time LUT-based color correction for cell phone cameras in low light conditions without frame memory

Low ambient light conditions usually result in under exposed images for cell phone cameras. This paper proposes a LUT-based histogram matching method to correct the color tone of the under-exposed images. This method can be performed in real-time and does not need frame memory.

Real-time look-up table-based color correction for still image stabilization of digital cameras without using frame memory

Abstract. Digital cameras usually decrease exposure time to capture motion-blur-free images. However, this operation will generate an under-exposed image with a low-budget complementary metal-oxide semiconductor image sensor (CIS). Conventional color correction algorithms can efficiently correct under-exposed images; however, they are generally not performed in real time and need at least one frame memory if they are implemented by hardware. The authors propose a real-time look-up table-based color correction method that corrects under-exposed images with hardware without using frame memory. The method utilizes histogram matching of two preview images, which are exposed for a long and short time, respectively, to construct an improved look-up table (ILUT) and then corrects the captured under-exposed image in real time. Because the ILUT is calculated in real time before processing the captured image, this method does not require frame memory to buffer image data, and therefore can greatly save the cost of CIS. This method not only supports single image capture, but also bracketing to capture three images at a time. The proposed method was implemented by hardware description language and verified by a field-programmable gate array with a 5 M CIS. Simulations show that the system can perform in real time with a low cost and can correct the color of under-exposed images well.

Low-Cost Panorama Unrolling of Catadioptric Omnidirectional Images

Catadioptric omni directional imaging systems generate images of 360 degrees field of view. The omni directional images need to be transformed into rectangular panorama images, because of the concentric circular distortion. In most application scenarios, it has to perform the panorama unrolling in real-time and at low-cost, especially for high-resolution images. In this paper, a low-cost hardware design is proposed, which adopts a software/hardware cooperative structure and employs several optimization schemes. To avoid the on-line division operation caused by the coordinate transformation algorithm, it is pre-computed and stored in the form of a look-up-table (LUT). And then, the amount of memory used by the LUT is reduced to 1/4 by using symmetrical characteristic. Experimental results show that the proposed hardware structure achieves an effective real-time performance and lower implementation cost, and it can be applied to other kinds of catadioptric omni directional imaging systems.