Motion detection in JPEG domain
Also available in printed version : TK7882.M68 L44 2014 raf
| 第一著者: | |
|---|---|
| その他の著者: | |
| フォーマット: | Master's thesis |
| 言語: | 英語 |
| 出版事項: |
Universiti Teknologi Malaysia
2025
|
| 主題: | |
| オンライン･アクセス: | https://utmik.utm.my/handle/123456789/114536 |
| Abstract | Abstract here |
| _version_ | 1854975102761828352 |
|---|---|
| author | Lee, Chai Yee |
| author2 | Muhammad Mun'im Ahmad Zabidi, supervisor |
| author_facet | Muhammad Mun'im Ahmad Zabidi, supervisor Lee, Chai Yee |
| author_sort | Lee, Chai Yee |
| description | Also available in printed version : TK7882.M68 L44 2014 raf |
| format | Master's thesis |
| id | utm-123456789-114536 |
| institution | Universiti Teknologi Malaysia |
| language | English |
| publishDate | 2025 |
| publisher | Universiti Teknologi Malaysia |
| record_format | dspace |
| record_pdf | Abstract |
| spelling | utm-123456789-1145362025-08-21T10:16:59Z Motion detection in JPEG domain Lee, Chai Yee Muhammad Mun'im Ahmad Zabidi, supervisor Motion detectors Image compression Also available in printed version : TK7882.M68 L44 2014 raf Nowadays, video surveillance turns out to be an active research. Along with the advancement of technologies, many surveillance cameras store image in Motion JPEG (MJPEG) which consists of many JPEG images that are encoded separately. Conventional motion detection is normally in pixels-domain, and this might involve unnecessary full decompression. Conversely, motion detection directly in JPEG domain offers performance benefit, such as reduces processing time, lower computational cost, and decreases power consumption. Therefore, the objectives of this project is to develop a motion detection algorithm in JPEG domain. The fundamental concept is to inspect the DC coefficients between adjacent JPEG frames. Basically, a MJPEG video which has a size of 640×480 was acquired. Afterwards, its raw data which includes the frame header, scan header, and Huffman tables was retrieved. Entropy decoding was performed on each 8×8 block for a JPEG image to obtain DC and AC coefficients. Next, the obtained coefficients were rearranged in zig-zag pattern. The same procedure occurred for successive JPEG frames. A comparison between information content on DC coefficients between two adjacent JPEG frames is conducted. If the absolute difference between the coefficients is above the threshold set, a motion is detected. In addition, a second threshold was set to define the total number of 8×8 blocks exceeding the DC difference value. Finally, a comparison in terms of accuracy for performing motion detection on partially decompressed JPEG frames versus fully decompressed was carried out. It was found out that out of 30 cases of tested data, in which a total of 60 JPEG images were involved, a 100% accuracy is achieved by pixels domain motion detection. Meanwhile, for JPEG domain motion detection, there were two cases of missed detection atiff UTM 133 p. Thesis (Sarjana Kejuruteraan (Elektrik - Elektronik dan Telekomunikasi)) – Universiti Teknologi Malaysia, 2014 2025-04-18T07:50:34Z 2025-04-18T07:50:34Z 2014-01 Master's thesis https://utmik.utm.my/handle/123456789/114536 valet-20150628-14258 vital:78026 ENG Closed Access UTM Complete Completion Unpublished application/pdf Universiti Teknologi Malaysia |
| spellingShingle | Motion detectors Image compression Lee, Chai Yee Motion detection in JPEG domain |
| thesis_level | Master |
| title | Motion detection in JPEG domain |
| title_full | Motion detection in JPEG domain |
| title_fullStr | Motion detection in JPEG domain |
| title_full_unstemmed | Motion detection in JPEG domain |
| title_short | Motion detection in JPEG domain |
| title_sort | motion detection in jpeg domain |
| topic | Motion detectors Image compression |
| url | https://utmik.utm.my/handle/123456789/114536 |
| work_keys_str_mv | AT leechaiyee motiondetectioninjpegdomain |