A Real-time Front Vehicle Distance and Lane Departure Warning System

• Main Authors: CHANG, SHUN-KAI, 張舜凱
• Other Authors: HUANG, JR-JEN
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/4eng4z

碩士 === 明志科技大學 === 電子工程系碩士班 === 106 === In this thesis, a real-time TI-DSP solution for front vehicle detection and lane departure warning system is proposed. The image processing algorithms are developed based on x86 platform and then implemented on a TI-DM6437 platform for a real-time application. The system process is starting with the settings of region of interest (ROI) in the image captured by the CCD lens, then the nearest neighbor interpolation is performed to reduce the amount of calculation afterwards. The noise is removed by the image processing algorithm, dynamic binarization and blob analysis are used to identify the shadow of front vehicle. When the shadow ratio of the under-base shade is within the threshold of the range, and the distance will be calculated. A lane dash line filled algorithm is used to enhance line feature. Peaking finding, and Hough transform are applied to extract the lane lines. The tracking will constantly update the current estimate value through actual data. When the vehicle deviates from the lane or is too close to the front vehicle, a warning message is displayed on the screen with an alarm to remind the driver.